Browsing by Author "Zhang, Yun"

Now showing 1 - 20 of 22
  • Thumbnail Image
    Item
    3D image processing and online mapping
    (University of New Brunswick, 2008) Foster, Burns; Zhang, Yun
  • Thumbnail Image
    Item
    3D information supported urban change detection using multi-angle and multi-sensor imagery
    (University of New Brunswick, 2015) Jabari, Shabnam; Zhang, Yun
    This PhD research is focused on urban change detection using very high resolution (VHR) imagery acquired by different sensors (i.e. airborne and satellite sensors) and different view angles. Thanks to high amount of details provided in VHR images, urban change detection is made possible. On the other hand, due to the complicated structure of 3D urban environments when projected into the 2D image spaces, detection of changes becomes complicated. In general, change detection is divided into two major steps: I. Establishment of a relation between bi-temporal images so that the corresponding pixels/segment are related; this is called co-registration; II. Comparison of the spectral properties of the co-registered pixels/segment in the bi-temporal images in order to detect changes. As far as Step 1 is concerned, establishment of an accurate global co-registration between bi-temporal images acquired by the different sensors is not possible in urban environments due to different geometric distortions in the imagery. Therefore, the majority of studies in this field avoid using multi-sensor and multi-view angle images. In this study, a novel co-registration method called "patch-wise co-registration" is proposed to address this problem. This method integrates the sensor model parameters into the co-registration process to relate the corresponding pixels and, by extension, the segments (patches). In Step 2, the brightness values of the matching pixels/segments are compared in order to detect changes. Thus, variations in the brightness values of the pixels/segments identify the changes. However, there are other factors that cause variations in the brightness values of the patches. One of them is the difference of the solar illumination angles in the bi-temporal images. In urban environment, the shape of the objects such as houses with steeply-sloped roofs (steep roofs) cause difference in the solar illumination angle resulting in difference in the brightness values of the associated pixels. This effect is corrected using irradiance topographic correction methods. Finally, the corrected irradiance of the co-registered patches is compared to detect changes using Multivariate Alteration Detection (MAD) transform. Generally, in the last stage of change detection process, "from-to" information is produced by checking the classification labels of the pixels/segments (patches). In this study, a fuzzy rule-based image classification methodology is proposed to improve the classification results, compared to the crisp thresholds, and accordingly increase the change detection accuracy. In total, the key results achieved in this research are: I. Including the off-nadir images and airborne images as the bi-temporal combinations in change detection; II. Solving the issue of geometric distortions in image co-registration step, caused by various looking angles of images, by introducing the patch-wise co-registration; III. Combining a robust spectral comparison method, which is the MAD transform, with the patch-wise change detection; IV. Removing the effect of illumination angle difference on the urban objects to improve change detection results; V. Improving classification results by using fuzzy thresholds in the image classification step. The outputs of this research provide an opportunity to utilize the huge amount of archived VHR imagery for automatic and semi-automatic change detection. Automatic classification of the images especially in urban area is still a challenge due to the spectral similarity between urban classes such as roads and buildings. Therefore, generation of the accurate “from-to” information is still remaining for future researches.
  • Thumbnail Image
    Item
    A scalable web tiled map management system
    (University of New Brunswick, 2017) Kotsollaris,, Menelaos; Stefanakis, Emmanuel; Zhang, Yun
    Modern map visualizations are built using data structures for storing tile images, while their main concerns are to maximize efficiency and usability. The core functionality of a web tiled map management system is to provide tile images to the end user; several tiles combined construe the web map. This thesis presents a comprehensive end-to-end analysis for developing and testing scalable web tiled map management systems. To achieve this, several data structures are showcased and analyzed. Specifically, this thesis focuses on the SimpleFormat, which stores the tiles directly on the file system; the ImageBlock, which divides each tile folder (a folder where the tile images are stored) into subfolders that contain multiple tiles prior to storing the tiles on the file system; the LevelFilesSet, a data structure that creates dedicated Random-Access files, wherein the tile dataset is first stored and then parsed in files to retrieve the tile images; and, finally, the LevelFilesBlock, a hybrid data structure which combines ImageBlock and LevelFilesSet data structures. This work signifies the first time this hybrid approach has been implemented and applied in a web tiled map context. Specifically, each data structure was implemented in Java. The JDBC API was used for integrating with the PostgreSQL database. This database was then used to conduct cross-testing amongst the data structures. Subsequently, several benchmark tests on local and cloud environments are developed anew and assessed under different system configurations to compare the data structures and provide a thorough analysis of their efficiency. These benchmarks showcased the efficiency of LevelFilesSet, which retrieved tiles up to 3.3 times faster than the other data structures. Peripheral features and principles of implementing scalable web tiled map management systems among different software architectures and system configurations are analyzed and discussed.
  • Thumbnail Image
    Item
    Application of footstep sound and lab colour space in moving object detection and image quality measurement using opposite colour pairs
    (University of New Brunswick, 2019) Roshan, Aditya; Zhang, Yun
    This PhD dissertation is focused on two of the major tasks of an Atlantic Innovation Fund (AIF) sponsored “Triple-sensitive Camera Project”. The first task focuses on the improvement of moving object detection techniques, and second on the evaluation of camera image quality. Cameras are widely used in security, surveillance, site monitoring, traffic, military, robotics, and other applications, where detection of moving objects is critical and important. Information about image quality is essential in moving object detection. Therefore, detection of moving objects and quality evaluation of camera images are two of the critical and challenging tasks of the AIF Triple-sensitive Camera Project. In moving object detection, frame-based and background-based are two major techniques that use a video as a data source. Frame-based techniques use two or more consecutive image frames to detect moving objects, but they only detect the boundaries of moving objects. Background-based techniques use a static background that needs to be updated in order to compensate for light change in a camera scene. Many background modelling techniques involving complex models are available which make the entire procedure very sophisticated and time consuming. In addition, moving object detection techniques need to find a threshold to extract a moving object. Different thresholding methodologies generate varying threshold values which also affect the results of moving object detection. When it comes to quality evaluation of colour images, existing Full-Reference methods need a perfect colour image as reference and No-Reference methods use a gray image generated from the colour image to compute image quality. However, it is very challenging to find a perfect reference colour image. When a colour image is converted to a grey image for image quality evaluation, neither colour information nor human colour perception is available for evaluation. As a result, different methods give varying quality outputs of an image and it becomes very challenging to evaluate the quality of colour images based on human vision. In this research, a single moving object detection using frame differencing technique is improved using footstep sound which is produced by the moving object present in camera scene, and background subtraction technique is improved by using opposite colour pairs of Lab colour space and implementing spatial correlation based thresholding techniques. Novel thresholding methodologies consider spatial distribution of pixels in addition to the statistical distribution used by existing methods. Out of four videos captured under different scene conditions used to measure improvements, a specified frame differencing technique shows an improvement of 52% in object detection rate when footstep sound is considered. Other frame-based techniques using Optical flow and Wavelet transform such are also improved by incorporating footstep sound. The background subtraction technique produces better outputs in terms of completeness of a moving object when opposite colour pairs are used with thresholding using spatial autocorrelation techniques. The developed technique outperformed background subtraction techniques with most commonly used thresholding methodologies. For image quality evaluation, a new “No-Reference” image quality measurement technique is developed which evaluates quantitative image quality score as it is evaluated by human eyes. The SCORPIQ technique developed in this research is independent of a reference image, image statistics, and image distortions. Colour segments of an image are spatially analysed using the colour information available in Lab colour space. Quality scores from SCORPIQ technique using LIVE image database yield distinguished results as compared to quality scores from existing methods which give similar results for visually different images. Compared to visual quality scores available with LIVE database, the quality scores from SCORPIQ technique are 3 times more distunquishable. SCORPIQ give 4 to 20 times distinguishable results compared to statistics based results which also does not follow the quality scores as evaluated by human eyes.
  • Thumbnail Image
    Item
    Building detection in off-nidar very high resolution satellite images based on stereo 3D information
    (University of New Brunswick, 2017) Suliman, Alaelidn Muhmud Housat; Zhang, Yun
    Mapping or updating maps of urban areas is crucial for urban planning and management. Since buildings are the main objects in urban environments, building roof detection is an important task in urban mapping. The ideal geo-spatial data source for mapping building information is very high resolution (VHR) satellite images. On the other hand, because buildings are elevated objects, incorporating their heights in building detection can significantly improve the accuracy of the mapping. The most cost-effective source for extracting the height information is stereo VHR satellite images that can provide two types of stereo 3D information: elevation and disparity. However, most VHR images are acquired off-nadir. This acquisition type causes building leaning in the images and creates major challenges for the incorporation of building height information into roof detection. Thus, this PhD research focuses on finding solutions to mitigate the problems associated with 3D-supported building detection in off-nadir VHR satellite images. It also exploits the potential of extracting disparity information from off-nadir image pairs to support building detection. In the research, several problems associated with building leaning need to be solved, such as building roof offsetting from its footprint, object occlusion, and building façades. Moreover, the variation of the roofs offsets based on the building heights. While the offsets of building roof create difficulties in the co-registration between image and elevation data, the building façades and occlusions create challenges in automatically finding matching points in off-nadir image pairs. Furthermore, due to the variation in building-roof offsets, the mapped roofs extracted from off-nadir images cannot be directly geo-referenced to existing maps for effective information integration. In this PhD dissertation, all of the above identified problems are addressed in a progressively improving manner (i.e., solving the problems one after another while improving the efficiency) within the context of 3D-supported building detection in off-nadir VHR satellite images. Firstly, an image-elevation co-registration technique is developed that is more efficient than the currently available techniques. Secondly, the computation cost is then reduced by generating disparity information instead of the traditional elevation data. This allows bypassing a few time-consuming steps of the traditional method. Thirdly, the disparity generation is then extended from using one pair of off-nadir images to using multiple pairs for achieving an enriched disparity map. Finally, the enriched disparity maps achieved are then used to efficiently derive elevations that are directly co-registered with pixel-level accuracy to the selected reference image. Based on these disparity-based co-registered elevations, building roofs are successfully detected and accurately geo-referenced to existing maps. The outcome of this PhD research proved the possibility of using off-nadir VHR satellite images for accurate urban building detection. It significantly increases the data source scope for building detection since most (> 95%) of VHR satellite images are off-nadir and traditional methods cannot effectively handle off-nadir images.
  • Thumbnail Image
    Item
    Delineation of Individual Tree Crowns Using High Spatial Resolution Multispectral WorldView-3 Satellite Imagery
    (Institute of Electrical and Electronics Engineers, 2021) Tong, Fei; Tong, Hengjian; Mishra, Rakesh; Zhang, Yun
    Delineation of individual tree crowns can provide valuable information for sustainable forest management and environmental protection. However, it is hard to find a reliable tree crown delineation method that can continually generate expected results in high spatial resolution multispectral satellite images, because most of the existing methods need user-assigned parameters that greatly affect the quality of the delineation results. In this article, we propose a method based on the marker-controlled watershed segmentation to delineate individual tree crowns using high spatial resolution multispectral WorldView-3 satellite imagery. A gradient binarization process is proposed to accurately locate tree crown borders. The threshold for the binarization is determined by a supervised searching process. Markers used in marker-controlled watershed segmentation are spatial local maxima detected from the information provided by tree crown borders. Moreover, the definition of spatial local maxima from the literature is improved to eliminate false treetops. To validate the performance of the proposed delineation method, delineation results are compared with those obtained from the spectral angle segmentation (SAS) method that has been proposed in literature because the quality of delineation results generated by SAS does not rely on user-assigned parameters. The experiment results in two test images demonstrate that the proposed method outperforms SAS in terms of both delineation accuracy and visual quality of the delineation map. Moreover, it is proved that the modified spatial local maxima are more reliable for detecting treetops.
  • Thumbnail Image
    Item
    Developing a deep learning network suitable for automated classification of heterogeneous land covers in high spatial resolution imagery
    (University of New Brunswick, 2019) Rezaee, Mohammad; Zhang, Yun
    The incorporation of spatial and spectral information within multispectral satellite images is the key for accurate land cover mapping, specifically for discrimination of heterogeneous land covers. Traditional methods only use basic features, either spatial features (e.g. edges or gradients) or spectral features (e.g. mean value of Digital Numbers or Normalized Difference Vegetation Index (NDVI)) for land cover classification. These features are called low level features and are generated manually (through so-called feature engineering). Since feature engineering is manual, the design of proper features is time-consuming, only low-level features in the information hierarchy can usually be extracted, and the feature extraction is application-based (i.e., different applications need to extract different features). In contrast to traditional land-cover classification methods, Deep Learning (DL), adapting the artificial neural network (ANN) into a deep structure, can automatically generate the necessary high-level features for improving classification without being limited to low-level features. The higher-level features (e.g. complex shapes and textures) can be generated by combining low-level features through different level of processing. However, despite recent advances of DL for various computer vision tasks, especially for convolutional neural networks (CNNs) models, the potential of using DL for land-cover classification of multispectral remote sensing (RS) images have not yet been thoroughly explored. The main reason is that a DL network needs to be trained using a huge number of images from a large scale of datasets. Such training datasets are not usually available in RS. The only few available training datasets are either for object detection in an urban area, or for scene labeling. In addition, the available datasets are mostly used for land-cover classification based on spatial features. Therefore, the incorporation of the spectral and spatial features has not been studied comprehensively yet. This PhD research aims to mitigate challenges in using DL for RS land cover mapping/object detection by (1) decreasing the dependency of DL to the large training datasets, (2) adapting and improving the efficiency and accuracy of deep CNNs for heterogeneous classification, (3) incorporating all of the spectral bands in satellite multispectral images into the processing, and (4) designing a specific CNN network that can be used for a faster and more accurate detection of heterogeneous land covers with fewer amount of training datasets. The new developments are evaluated in two case studies, i.e. wetland detection and tree species detection, where high resolution multispectral satellite images are used. Such land-cover classifications are considered as challenging tasks in the literature. The results show that our new solution works reliably under a wide variety of conditions. Furthermore, we are releasing the two large-scale wetland and tree species detection datasets to the public in order to facilitate future research, and to compare with other methods.
  • Thumbnail Image
    Item
    Enhanced Gaussian background modeling algorithm and implementation in FPGA for real-time moving object detection in surveillance video
    (University of New Brunswick, 2014) Guo, Ge; Zhang, Yun; Kaye, Mary
    A real-time solution of moving object detection (MOD) in surveillance video was explored in this work motivated by the practical ileed of real-time automated video analysis system. The core element of a moving object detection process is its background modeling algorithm in the content of surveillance and road monitoring applications. By reviewing and analyzing previous works, single Gaussian (SG) background modeling algorithm was selected and enhanced. Then a circuit that performs MOD with enhanced SG algorithm was designed and implemented in a Virtex6 FPGA of a ML605 evaluation board with other hardware components. The experiment results showed that the proposed MOD system could perform real-time MOD in a video of 1280x720p@30fps. It outperforms the software experiments/implementations and the state-of-art FPGA-based implementations.
  • Thumbnail Image
    Item
    Forestry information extraction using high spatial resolution remote sensing imagery
    (University of New Brunswick, 2023-12) Tong, Fei; Zhang, Yun
    This PhD research focuses on the development of reliable and efficient methods for individual tree crown delineation and tree species classification, which provide essential information for modern forestry management and climate change monitoring. The primary objective of this research is to develop robust methods that can accurately delineate individual tree crowns and classify tree species using high spatial resolution remote sensing imagery. By achieving this objective, the research aims to enhance the reliability and efficiency of forestry management practices and contribute to the field of remote sensing applications in forestry. For tree crown delineation task, existing tree crown delineation methods are not suitable for large areas applications, because they need highly experienced experts to manually assign suitable parameters to control the delineation results, which is time-consuming, inaccurate, and not suitable for normal users. To address this issue, this dissertation presents a tree crown delineation method utilizing marker-controlled watershed segmentation specifically designed for high spatial resolution multispectral WorldView-3 satellite imagery. To reduce the difficulty of assigning parameters, the proposed method incorporates an automated supervised search process to determine the threshold. Moreover, an enhanced definition of spatial local maximum is employed to mitigate false treetops, thereby enhancing the accuracy of treetop detection. For tree species classification task, although deep learning methods based on convolutional neural networks (CNN) have achieved promising results, challenges remain in hyperparameter tuning and the requirement for large number of labeled training samples, limiting their applicability in real-world applications. This dissertation addresses these challenges by proposing three models based on the concept of deep forest to enhance the tree species classification from high spatial resolution hyperspectral imagery. All the three proposed models only require two hyperparameters that are easy to be determined by users. To optimize the classification accuracy, two different ways to combine both fixed-size patches and shape-adaptive superpixels are proposed to fully exploit spectral-spatial information within the high spatial resolution hyperspectral imagery. To reduce the demand for labeled training samples, the active learning (AL) is perfectly integrated into the multilayer cascaded random forests classification model.
  • Thumbnail Image
    Item
    Hyperspectral remote sensing in lithological mapping, mineral exploration, and environmental geology: an updated review
    (Society of Photo-optical Instrumentation Engineers, 2021) Peyghambari, Sima; Zhang, Yun
    Hyperspectral imaging has been used in a variety of geological applications since its advent in the 1970s. In the last few decades, different techniques have been developed by geologists to analyze hyperspectral data in order to quantitatively extract geological information from the high-spectral-resolution remote sensing images. We attempt to review and update various steps of the techniques used in geological information extraction, such as lithological and mineralogical mapping, ore exploration, and environmental geology. The steps include atmospheric correction, dimensionality processing, endmember extraction, and image classification. It is identified that per-pixel and subpixel image classifiers can generate accurate alteration mineral maps. Producing geological maps of different surface materials including minerals and rocks is one of the most important geological applications. The hyperspectral images classification methods demonstrate the potential for being used as a main tool in the mining industry and environmental geology. To exemplify the potential, we also include a few case studies of different geological applications.
  • Thumbnail Image
    Item
    Implementing scalable geoweb applications using cloud and internet computing
    (University of New Brunswick, 2014) Mousavi, Seyed; Zhang, Yun; Wachowicz, Monica
    New advancements in technology such as the rise of social networks have led to more geospatial data being produced every day. The current issue with the large volume of geospatial data is to store and process it because of the scalability of the data.In this thesis, two computing implementations, cloud computing and Internet computing, are studied and evaluated for their capability in storing,processing and visualizing large volumes of geospatial data. For the cloud computing implementation,the different concepts of cloud computing have been analysed according to their applications, models and services. Moreover, a case study using cloud computing platforms has also been implemented for storing and processing geotagged tweets retrieved for a national recreational park in Vancouver, BC. For the Internet computing platform,the Open Geospatial Consortium’s Web Processing Service has been investigated as a framework for sharing geospatial data and processing it over Internet. A raster calculation algorithm in Web Processing Service platforms has also been implemented on 2 scenes of Lands at satellite imagery to evaluate WPS’ capabilities in handling large volume of data. Results of this research suggest that internet computing can be used to handle geospatial data processing but,when dealing with large volumes of data,this study proves that Internet computing and current Geospatial Information Systems are not suitable to be used and cloud computing platform can be utilized to handle large volumes of geospatial data.
  • Thumbnail Image
    Item
    Improving forward mapping and disocclusion inpainting algorithms for depth-image-based rendering and geomatics applications
    (University of New Brunswick, 2022-08) Liu, Weilong (William); Zhang, Yun
    Depth-image-based Rendering (DIBR) is a promising technique to generate appealing and immersive contents for virtual reality image mapping. When mapping a reference image to a virtual image, two challenges exist: (1) the most widely used forward mapping algorithm is prone to generating synthetic artifacts; and (2) the existing disocclusion inpainting algorithms are not able to fill big holes in a visually plausible way. This research conducted a comprehensive investigation on the forward mapping algorithm and developed a new algorithm that can eliminate all the reported artifacts of small cracks. This research also developed a novel directional and adaptive hole-filling algorithm using an exemplar-based approach to fill big holes. The performance of the developed algorithms was validated using images from a selected benchmark database. Quantitative and qualitative assessments proved that the developed algorithms have clearly outperformed the state-of-the-art algorithms. The developed algorithms were applied to outdoor/indoor mapping applications to prove the concept. It validated the feasibility of using the developed algorithms to improve user experiences on existing Street View technologies.
  • Thumbnail Image
    Item
    Improving spatial quality of terrestrial and satellite images by demosaicking and fusion
    (University of New Brunswick, 2022-05) Fathollahikalanpa, Fatemeh; Zhang, Yun
    Improving the spatial quality of a colour image brings valuable benefits to all imaginable applications of the image. One method for such an improvement is to incorporate ‘panchromatic’ sensors into imaging. Panchromatic sensors provide images with higher spatial quality than colour images because they do not filter any complementary colours of the incoming light. Combining panchromatic and colour sensors has been employed in different fields. In remote sensing (RS), panchromatic and multispectral / hyperspectral images are captured by two separate sensor chips and then fused through pan-sharpening techniques. In terrestrial applications, a single sensor chip is used to accommodate both panchromatic (or white, W) and colour (RGB) pixels using a Colour Filter Array (RGBW CFA). A ‘demosaicking’ procedure needs to be employed to generate RGB colour images. Both pan-sharpening and RGBW demosaicking still have unsolved problems despite being used by the imaging industry for a while. In RS, most hyperspectral bands are not pan-sharpened, because they fall beyond the panchromatic spectral range, causing significant spectral distortion. For RGBW demosaicking, limited methods have been published which produce low-quality images, mainly because they demosaick panchromatic and colour images independently. Another issue is that existing approaches cannot handle images corrupted by noise, because they do not involve denoising. This dissertation aims to overcome the above-mentioned obstacles in improving the spatial quality of the hyperspectral/colour images. For hyperspectral images, this research develops an adaptive band selection strategy to identify the bands across the entire spectrum that can be pan-sharpened without introducing high spectral distortion. For RGBW demosaicking, this research firstly proposes a collaborative interpolation between panchromatic and colour pixels. It significantly improves the spatial quality by reducing the zipper effects and retaining the spatial details. The research then proposes a deep learning-based approach for RGBW joint demosaicking and denoising, along with a procedure to prepare the required training dataset. Results show a considerable quality improvement over existing methods even for images corrupted by various noise levels. In summary, this research leads to improving the spatial quality of those hyperspectral bands, that were previously left unfused. It also increases the potential of using RGBW cameras in daily applications due to the significant quality boost.
  • Thumbnail Image
    Item
    Multigranularity Multiclass-Layer Markov Random Field Model for Semantic Segmentation of Remote Sensing Images
    (IEEE, 2020) Zheng, Chen; Zhang, Yun; Wang, Leiguang
    Semantic segmentation is one of the most important tasks in remote sensing. However, as spatial resolution increases, distinguishing the homogeneity of each land class and the heterogeneity between different land classes are challenging. The Markov random field model (MRF) is a widely used method for semantic segmentation due to its effective spatial context description. To improve segmentation accuracy, some MRF-based methods extract more image information by constructing the probability graph with pixel or object granularity units, and some other methods interpret the image from different semantic perspectives by building multilayer semantic classes. However, these MRF-based methods fail to capture the relationship between different granularity features extracted from the image and hierarchical semantic classes that need to be interpreted. In this article, a new MRF-based method is proposed to incorporate the multigranularity information and the multilayer semantic classes together for semantic segmentation of remote sensing images. The proposed method develops a framework that builds a hybrid probability graph on both pixel and object granularities and defines a multiclass-layer label field with hierarchical semantic over the hybrid probability graph. A generative alternating granularity inference is suggested to provide the result by iteratively passing and updating information between different granularities and hierarchical semantics. The proposed method is tested on texture images, different remote sensing images obtained by the SPOT5, Gaofen-2, GeoEye, and aerial sensors, and Pavia University hyperspectral image. Experiments demonstrate that the proposed method shows a better segmentation performance than other state-of-the-art methods.
  • Thumbnail Image
    Item
    Optimization of the shoreline delineation process for Canada's coastlines
    (University of New Brunswick, 2018) Callaghan, Riley; Baker, Jessica; Hofland, Rebekah; McLennan, Mary; Fraser, David; Zhang, Yun
  • Thumbnail Image
    Item
    Remote sensing of spruce budworm defoliation in Quebec, Canada using EO-1 Hyperion data
    (University of New Brunswick, 2015) Huang, Zhongwei; Zhang, Yun
    Satellite remote sensing has special advantages for monitoring the extent of defoliation caused by insects. Remote sensing has been used to monitor spruce budworm defoliation, mostly using the data captured by multispectral sensors such as Landsat (MSS, TM, and ETM+), MODIS and SPOT. However, these images have a low spectral resolution (using 4 to 36 spectral bands each covers a broad spectral bandwidth) which limited their abilities to identify small spectral variations in individual pixels for diagnosing specific forest insect infection. As an alternative, the hyperspectral data provided by EO-1 Hyperion sensor provides a high spectral resolution using 242 spectral bands from 0.4 to 2.5 ìm (each band covers a very narrow spectral bandwidth). However, little study has been conducted on using Hyperion or other satellite hyperspectral images for monitoring spruce budworm defoliation. Taking advantage of the rich spectral information, this thesis proposed methods for remotely sensing, estimating and mapping spruce budworm (SBW) defoliation using the spectral information, i.e. vegetation indices (VIs), extracted from Hyperion images. 15% of accuracy improvement in SBW defoliation estimation and mapping has been achieved by applying the developed Hyperion VIs compared with conventional multispectral VIs. Highly accurate mapping results have been generated by developing suitable feature extraction method.
  • Thumbnail Image
    Item
    Solar photovoltaic panel and roofing material detection using WorldView-3 imagery
    (University of New Brunswick, 2017) Mishra, Rakesh Kumar; Zhang, Yun
    This PhD dissertation focuses on the development of new techniques to detect urban solar photovoltaic (PV) panel installations and roofing materials utilizing the commercially available WorldView-3 satellite imagery, consisting of 1 panchromatic (Pan) band with 0.3m resolution, 8 visible and near infrared (VNIR) bands with 1.2m resolution, and 8 short wave infrared (SWIR) spectral bands with 7.5m resolution. To accurately detect urban solar PV panels and roofing materials, it is necessary to analyze the spectral information in both the 8 VNIR bands and the 8 SWIR bands at the pixel level. However, the resolution difference between the VNIR bands and the SWIR bands is more than 6 times, which creates significant challenges for the spectral analysis and thus for the material detection. In order to increase the resolution of the SWIR bands from 7.5m to 1.2m, Fuze Go SWIR Sharp (FGSS) algorithm is used. The resulting high-resolution 1.2m SWIR bands are then combined with the original 1.2m VNIR bands to form a 16-band 1.2m (VNIR+SWIR) super spectral imagery. A method to detect solar PV panel installations and a method to detect roofing materials in the 16-band super spectral imagery are also developed. In order to increase the resolution of WorldView-3 SWIR bands from 7.5m to 1.2m and take advantage of their capability for material identification, this research investigated the capacities of 9 popular, industry adopted pan-sharpening algorithms for pan-sharpening the WorldView-3 SWIR bands. The general principles of the pan-sharpening algorithms are reviewed. The WorldView-3 Pan images were down-sampled from 0.4m to 1.6m to fuse with the 7.5m SWIR image. Experiments demonstrate that the most commonly used algorithms are not suitable for pan-sharpening SWIR images, whereas the new pan-sharpening algorithm, Fuze Go SWIR Sharp (FGSS), can produce satisfactory results. The reasons why most algorithms fail to produce quality pan-sharpened SWIR bands are also examined. To detect solar PV panels, a new method is developed that can effectively analyze the spectral information in the newly formed high-resolution (HR) 16-band 1.2m super spectral (SS) imagery by adapting the spectral angle mapping (SAM) algorithm. The proposed method, named HR-SSF-SAM method, is tested on the WorldView-3 imagery of Brea, California, USA. The results demonstrate a true detection rate of 93.3% with 0% false detection. Even solar PV panels and glass roofs can be differentiated from each other. To detect roofing materials, such as fiberglass, ethylene propylene diene monomer (EPDM), metal, and concrete, using WorldView-3 imagery, a novel method is proposed. The method utilizes the newly formed high-resolution 16-band 1.2m super spectral imagery and introduces a new approach to detect roofing materials. Experiments with the WorldView-3 imagery of Brea, California, USA, demonstrate that the proposed method achieves an overall accuracy of 97.59% and Kappa accuracy of 95.59% for roofing material detection in commercial areas, and an overall accuracy of 93.88% and Kappa accuracy of 88.98% for roofing material detection in residential areas with family houses. Because of the complexity of using WorldView-3 imagery for solar PV panel detection and roofing material detection, very few publications can be found in this area. The literature review undertaken for this research confirms that the accuracies achieved are significantly better than those found in the literature.
  • Thumbnail Image
    Item
    Spatial analysis of land cover changes in the Grand Lake Meadows, New Brunswick
    (University of New Brunswick, 2016) Shodimu, Oluwatimilehin Okikiolu; Al-Tahir, Raid; Zhang, Yun
    The ever growing human activities and economic development will eventually change the relationships between human and the environment. A matter of grave concern is the unsustainable patterns of land use that are considered a major cause for the deterioration of the environment. The Grand Lake Meadows is an important part of the Saint John River wetlands that form the largest freshwater wetland habitat in the Maritimes (east Canada). In this paper, remotely sensed images were used for mapping the use of land use and cover in the Grand Lake Meadow over a period of 20 years. The goal was to undertake a detailed spatially explicit inventory of local trends in land use and land cover changes through classifying the historical images. Other available data like the road network to mention a few were combined with this information to create a database that was used to investigate consequences of land use/cover change. The results demonstrates the flexibility and effectiveness of this technology in establishing the necessary baseline and support information for sustaining eco-services of a wetland thereby depicting the rate of change undergone in the GLM area over time. The study identified a 38% decrease in the wetland from the 1990 to 2001, while there was 4.32% overall increase in the wetland area since then. The result will help the managers to comprehend the dynamics of the changes, prompting a better management and implementation of LULC administration in the GLM area.
  • Thumbnail Image
    Item
    Surface water quality assessment using a remote sensing, GIS, and mathematical modelling framework
    (University of New Brunswick, 2018) Sharaf El Din, Essam; Zhang, Yun
    The presence of various pollutants in water bodies can lead to the deterioration of both surface water quality and aquatic life. Surface water quality researchers are confronted with significant challenges to properly assess surface water quality in order to provide an appropriate treatment to water bodies in a cost-effective manner. Conventional surface water quality assessment methods are widely performed using laboratory analysis, which are labour intensive, costly, and time consuming. Moreover, these methods can only provide individual concentrations of surface water quality parameters (SWQPs), measured at monitoring stations and shown in a discrete point format, which are difficult for decision-makers to understand without providing the overall patterns of surface water quality. In contrast, remote sensing has shown significant benefits over conventional methods because of its low cost, spatial continuity, and temporal consistency. Thus, exploring the potential of using remotely sensed data for surface water quality assessment is important for improving the efficiency of surface water quality evaluation and water body treatment. In order to properly assess surface water quality from satellite imagery, the relationship between satellite multi-spectral data and concentrations of SWQPs should be modelled. Moreover, to make the process accessible to decision-makers, it is important to extract the accurate surface water quality levels from surface water quality raw data. Additionally, to improve the cost effectiveness of surface water body treatment, identifying the major pollution sources (i.e., SWQPs) that negatively influence water bodies is essential. Therefore, this PhD dissertation focuses on the development of new techniques for (1) estimating the concentrations of both optical and non-optical SWQPs from a recently launched earth observation satellite (i.e., Landsat 8), which is freely available and has the potential to support coastal studies, (2) mapping the complex relationship between satellite multi-spectral signatures and concentrations of SWQPs, (3) simplifying the expression of surface water quality and delineating the accurate levels of surface water quality in water bodies, and (4) classifying the most significant SWQPs that contribute to both spatial and temporal variations of surface water quality. The outcome of this PhD dissertation proved the feasibility of developing models to retrieve the concentrations of both optical and non-optical SWQPs from satellite imagery with highly accurate estimations. It exhibited the potential of using remote sensing to achieve routine water quality monitoring. Moreover, this research demonstrated the possibility of improving the accuracy of surface water quality level extraction with inexpensive implementation cost. Finally, this research showed the capability of using satellite data to provide continuously updated information about surface water quality, which can support the process of water body treatment and lead to effective savings and proper utilization of surface water resources.
  • Thumbnail Image
    Item
    Terrestial implementation of UNB Super Camera and improvements to UNB-PanSharp
    (University of New Brunswick, 2015) Adham Khiabani, Sina; Zhang, Yun
    Camera sensitivity is a significant challenge for many imaging applications, especially in low light conditions. Image recognition and presentation in low light conditions is highly dependent on camera sensitivity. Issues acquiring colour images in low light conditions are amplified because of the fact that the colour images are acquired in a narrow spectral band. To address this issue it is possible to collect images in black and white (monochrome). The wide spectral coverage of such monochrome cameras can improve the sensitivity of the resulting images with the same sensors, but the colour will be sacrificed in this strategy. Another solution would be to use lower resolution colour cameras to increase the signal-to-noise ratio. This solution will result in less spatial detail. In satellite systems, to improve the sensitivity of the images, a pair of high resolution monochrome and low resolution colour cameras is used. Fusion of the images from those cameras will result in a high sensitivity and high resolution colour image. This thesis investigates the potential to implement this technology in a terrestrial configuration, using a security camera application as an example. UNB Super Camera is a high resolution monochrome camera coupled with a lower resolution colour camera which, when processed using UNB PanSharp technique, produces high resolution colour video. In order to implement UNB Super Camera for a terrestrial application, a system with four components was designed (data collection, processing, display / storage and framework software). All of the components were researched and studied in this thesis with the results of this work being used as inputs into the design and development of a terrestrial based UNB Super Camera system. The data collection review included issues associated with the camera, and its associated hardware requirements. Data processing included frame-to-frame co-registration, photogrammetric calibration and orientation that facilitated image fusion, motion detection and tracking and 3D positioning. Data display / storage was facilitated with a standard monitor and computer storage facilities. The key component of the system design and implementation is the framework software which is .NET based and has been designed and developed to facilitate the real-time operation of the UNB Super Camera system. The system was been successfully implemented and the results obtained were assessed as to their quality using the criteria of sensitivity, resolution and colour rendering. It should be noted that while a complete UNB Super Camera system has been designed and implemented, ONLY the imagery components are addressed in detail in this thesis. The motion object detection / tracking / 3D positioning components, as required by a security camera application are not analyzed in detail. These subjects are the focus of other researchers. The results of the assessment proved that the UNB Super Camera had measurably higher sensitivity and resolution and colour rendering in comparison with the same generation of available high resolution colour cameras, especially in lower lighting conditions. Despite this improvement, the fused images / videos had colour distortions and stain in very low lighting indoor cases and sunshine condition in outdoor cases. Investigation into these issues showed that the different spectral coverage of the high resolution monochrome camera and low resolution colour camera was the source of the problems. To address the contaminations, four methods -- including Fixed Coefficient, Adaptive Component, Monochrome Correction and Differential Filtering -- were proposed and investigated. Implementation of these strategies showed that the differential filtering method provided the best results. However, all of the methods were successful in recovering the distortions and stains in different lighting conditions, to varying degrees. In addition, the sensitivity, resolution and colour rendering of the results were further improved. Beside the spectral coverage effects, a debayering issue has also appeared in this project. Debayering effects of the low resolution colour were inherited by the high resolution fused videos. To address this issue, a combined Gaussian debayering and binning strategy was proposed. Although the resulting debayered and binned video was slightly blurred in comparison with the original debayered and binned low resolution colour, the resulting fused video frames using this method led to measurably higher-quality frames. Moreover, this method was computationally faster in comparison with the other methods, which is important in real-time applications.