A Multi-layer Perceptron based Non-linear Mixture Model to estimate class abundance from mixed pixels

Sub-pixel classification is essential for the successful description of many land cover (LC) features with spatial resolution less than the size of the image pixels. A commonly used approach for sub-pixel classification is linear mixture models (LMM). Even though, LMM have shown acceptable results, pragmatically, linear mixtures do not exist. A non-linear mixture model, therefore, may better describe the resultant mixture spectra for endmember (pure pixel) distribution. In this paper, we propose a new methodology for inferring LC fractions by a process called automatic linear-nonlinear mixture model (AL-NLMM). AL-NLMM is a three step process where the endmembers are first derived from an automated algorithm. These endmembers are used by the LMM in the second step that provides abundance estimation in a linear fashion. Finally, the abundance values along with the training samples representing the actual proportions are fed to multi-layer perceptron (MLP) architecture as input to train the neurons which further refines the abundance estimates to account for the non-linear nature of the mixing classes of interest. AL-NLMM is validated on computer simulated hyperspectral data of 200 bands. Validation of the output showed overall RMSE of 0.0089±0.0022 with LMM and 0.0030±0.0001 with the MLP based AL-NLMM, when compared to actual class proportions indicating that individual class abundances obtained from AL-NLMM are very close to the real observations.

An approach to multi-temporal MODIS image analysis using image classification and segmentation

This paper discusses an approach for river mapping and flood evaluation based on multi-temporal time series analysis of satellite images utilizing pixel spectral information for image classification and region-based segmentation for extracting water-covered regions. Analysis of MODIS satellite images is applied in three stages: before flood, during flood and after flood. Water regions are extracted from the MODIS images using image classification (based on spectral information) and image segmentation (based on spatial information). Multi-temporal MODIS images from ``normal'' (non-flood) and flood time-periods are processed in two steps. In the first step, image classifiers such as Support Vector Machines (SVMs) and Artificial Neural Networks (ANNs) separate the image pixels into water and non-water groups based on their spectral features. The classified image is then segmented using spatial features of the water pixels to remove the misclassified water. From the results obtained, we evaluate the performance of the method and conclude that the use of image classification (SVM and ANN) and region-based image segmentation is an accurate and reliable approach for the extraction of water-covered regions. (c) 2012 COSPAR. Published by Elsevier Ltd. All rights reserved.

Integration of speckle de-noising and image segmentation using synthetic aperture radar image for flood extent extraction

Flood is one of the detrimental hydro-meteorological threats to mankind. This compels very efficient flood assessment models. In this paper, we propose remote sensing based flood assessment using Synthetic Aperture Radar (SAR) image because of its imperviousness to unfavourable weather conditions. However, they suffer from the speckle noise. Hence, the processing of SAR image is applied in two stages: speckle removal filters and image segmentation methods for flood mapping. The speckle noise has been reduced with the help of Lee, Frost and Gamma MAP filters. A performance comparison of these speckle removal filters is presented. From the results obtained, we deduce that the Gamma MAP is reliable. The selected Gamma MAP filtered image is segmented using Gray Level Co-occurrence Matrix (GLCM) and Mean Shift Segmentation (MSS). The GLCM is a texture analysis method that separates the image pixels into water and non-water groups based on their spectral feature whereas MSS is a gradient ascent method, here segmentation is carried out using spectral and spatial information. As test case, Kosi river flood is considered in our study. From the segmentation result of both these methods are comprehensively analysed and concluded that the MSS is efficient for flood mapping.

Multi-Temporal Satellite Image Analysis Using Gene Expression Programming

This paper discusses an approach for river mapping and flood evaluation to aid multi-temporal time series analysis of satellite images utilizing pixel spectral information for image classification and region-based segmentation to extract water covered region. Analysis of Moderate Resolution Imaging Spectroradiometer (MODIS) satellite images is applied in two stages: before flood and during flood. For these images the extraction of water region utilizes spectral information for image classification and spatial information for image segmentation. Multi-temporal MODIS images from ``normal'' (non-flood) and flood time-periods are processed in two steps. In the first step, image classifiers such as artificial neural networks and gene expression programming to separate the image pixels into water and non-water groups based on their spectral features. The classified image is then segmented using spatial features of the water pixels to remove the misclassified water region. From the results obtained, we evaluate the performance of the method and conclude that the use of image classification and region-based segmentation is an accurate and reliable for the extraction of water-covered region.

PolyMage: Automatic Optimization for Image Processing Pipelines

This paper presents the design and implementation of PolyMage, a domain-specific language and compiler for image processing pipelines. An image processing pipeline can be viewed as a graph of interconnected stages which process images successively. Each stage typically performs one of point-wise, stencil, reduction or data-dependent operations on image pixels. Individual stages in a pipeline typically exhibit abundant data parallelism that can be exploited with relative ease. However, the stages also require high memory bandwidth preventing effective utilization of parallelism available on modern architectures. For applications that demand high performance, the traditional options are to use optimized libraries like OpenCV or to optimize manually. While using libraries precludes optimization across library routines, manual optimization accounting for both parallelism and locality is very tedious. The focus of our system, PolyMage, is on automatically generating high-performance implementations of image processing pipelines expressed in a high-level declarative language. Our optimization approach primarily relies on the transformation and code generation capabilities of the polyhedral compiler framework. To the best of our knowledge, this is the first model-driven compiler for image processing pipelines that performs complex fusion, tiling, and storage optimization automatically. Experimental results on a modern multicore system show that the performance achieved by our automatic approach is up to 1.81x better than that achieved through manual tuning in Halide, a state-of-the-art language and compiler for image processing pipelines. For a camera raw image processing pipeline, our performance is comparable to that of a hand-tuned implementation.

A closely integrated digital image capture device for embedded applications

A modular image capture system with close integration to CCD cameras has been developed. The aim is to produce a system capable of integrating CCD sensor, image capture and image processing into a single compact unit. This close integration provides a direct mapping between CCD pixels and digital image pixels. The system has been interfaced to a digital signal processor board for the development and control of image processing tasks. These have included characterization and enhancement of noisy images from an intensified camera and measurement to subpixel resolutions. A highly compact form of the image capture system is in an advanced stage of development. This consists of a single FPGA device and a single VRAM providing a two chip image capturing system capable of being integrated into a CCD camera. A miniature compact PC has been developed using a novel modular interconnection technique, providing a processing unit in a three dimensional format highly suited to integration into a CCD camera unit. Work is under way to interface the compact capture system to the PC using this interconnection technique, combining CCD sensor, image capture and image processing into a single compact unit. ©2005 Copyright SPIE - The International Society for Optical Engineering.

Mean shift based gradient vector flow for image segmentation

In recent years, gradient vector flow (GVF) based algorithms have been successfully used to segment a variety of 2-D and 3-D imagery. However, due to the compromise of internal and external energy forces within the resulting partial differential equations, these methods may lead to biased segmentation results. In this paper, we propose MSGVF, a mean shift based GVF segmentation algorithm that can successfully locate the correct borders. MSGVF is developed so that when the contour reaches equilibrium, the various forces resulting from the different energy terms are balanced. In addition, the smoothness constraint of image pixels is kept so that over- or under-segmentation can be reduced. Experimental results on publicly accessible datasets of dermoscopic and optic disc images demonstrate that the proposed method effectively detects the borders of the objects of interest.

A Survey of Image-based Relighting Techniques

Image-based Relighting (IBRL) has recently attracted a lot of research interest for its ability to relight real objects or scenes, from novel illuminations captured in natural/synthetic environments. Complex lighting effects such as subsurface scattering, interreflection, shadowing, mesostructural self-occlusion, refraction and other relevant phenomena can be generated using IBRL. The main advantage of image-based graphics is that the rendering time is independent of scene complexity as the rendering is actually a process of manipulating image pixels, instead of simulating light transport. The goal of this paper is to provide a complete and systematic overview of the research in Imagebased Relighting. We observe that essentially all IBRL techniques can be broadly classified into three categories (Fig. 9), based on how the scene/illumination information is captured: Reflectance function-based, Basis function-based and Plenoptic function-based. We discuss the characteristics of each of these categories and their representative methods. We also discuss about the sampling density and types of light source(s), relevant issues of IBRL.

Hyperspectral Image Classification with Nonlinear Methods

This thesis introduces two related lines of study on classification of hyperspectral images with nonlinear methods. First, it describes a quantitative and systematic evaluation, by the author, of each major component in a pipeline for classifying hyperspectral images (HSI) developed earlier in a joint collaboration [23]. The pipeline, with novel use of nonlinear classification methods, has reached beyond the state of the art in classification accuracy on commonly used benchmarking HSI data [6], [13]. More importantly, it provides a clutter map, with respect to a predetermined set of classes, toward the real application situations where the image pixels not necessarily fall into a predetermined set of classes to be identified, detected or classified with.

The particular components evaluated are a) band selection with band-wise entropy spread, b) feature transformation with spatial filters and spectral expansion with derivatives c) graph spectral transformation via locally linear embedding for dimension reduction, and d) statistical ensemble for clutter detection. The quantitative evaluation of the pipeline verifies that these components are indispensable to high-accuracy classification.

Secondly, the work extends the HSI classification pipeline with a single HSI data cube to multiple HSI data cubes. Each cube, with feature variation, is to be classified of multiple classes. The main challenge is deriving the cube-wise classification from pixel-wise classification. The thesis presents the initial attempt to circumvent it, and discuss the potential for further improvement.

Three-dimensional medical ultrasound

A number of groups around the world are working in the field of three dimensional(3D) ultrasound (US) in order to obtain higher quality diagnostic information. 3D US, in general, involves collecting a sequence of conventional 2D US images along with information on the position and orientation of each image plane. A transformation matrix is calculated relating image space to real world space. This allows image pixels and region of interest (ROI) points drawn on the image to be displayed in 3D. The 3D data can be used for the production of volume or surface rendered images, or for the direct calculation of ROI volumes.

Contributions to the analysis and segmentation of remote sensing hyperspectral images

142 p.

Moving Object Scanning Apparatus and Method

Apparatus for scanning a moving object includes a visible waveband sensor oriented to collect a series of images of the object as it passes through a field of view. An image processor uses the series of images to form a composite image. The image processor stores image pixel data for a current image and predecessor image in the series. It uses information in the current image and its predecessor to analyse images and derive likelihood measures indicating probabilities that current image pixels correspond to parts of the object. The image processor estimates motion between the current image and its predecessor from likelihood weighted pixels. It generates the composite image from frames positioned according to respective estimates of object image motion. Image motion may alternatively be detected be a speed sensor such as Doppler radar sensing object motion directly and providing image timing signals

Active regions for colour texture segmentation integrating region and boundary information

In this paper a colour texture segmentation method, which unifies region and boundary information, is proposed. The algorithm uses a coarse detection of the perceptual (colour and texture) edges of the image to adequately place and initialise a set of active regions. Colour texture of regions is modelled by the conjunction of non-parametric techniques of kernel density estimation (which allow to estimate the colour behaviour) and classical co-occurrence matrix based texture features. Therefore, region information is defined and accurate boundary information can be extracted to guide the segmentation process. Regions concurrently compete for the image pixels in order to segment the whole image taking both information sources into account. Furthermore, experimental results are shown which prove the performance of the proposed method

Robust background model for pixel based people counting using a single uncalibrated camera

Several pixel-based people counting methods have been developed over the years. Among these the product of scale-weighted pixel sums and a linear correlation coefficient is a popular people counting approach. However most approaches have paid little attention to resolving the true background and instead take all foreground pixels into account. With large crowds moving at varying speeds and with the presence of other moving objects such as vehicles this approach is prone to problems. In this paper we present a method which concentrates on determining the true-foreground, i.e. human-image pixels only. To do this we have proposed, implemented and comparatively evaluated a human detection layer to make people counting more robust in the presence of noise and lack of empty background sequences. We show the effect of combining human detection with a pixel-map based algorithm to i) count only human-classified pixels and ii) prevent foreground pixels belonging to humans from being absorbed into the background model. We evaluate the performance of this approach on the PETS 2009 dataset using various configurations of the proposed methods. Our evaluation demonstrates that the basic benchmark method we implemented can achieve an accuracy of up to 87% on sequence ¿S1.L1 13-57 View 001¿ and our proposed approach can achieve up to 82% on sequence ¿S1.L3 14-33 View 001¿ where the crowd stops and the benchmark accuracy falls to 64%.

Gamma Knife(A (R)) 3-D Dose Distribution Mapping by Magnetic Resonance Imaging

In medical processes where ionizing radiation is used, dose planning and dose delivery are the key elements to patient safety and treatment success, particularly, when the delivered dose in a single session of treatment can be an order of magnitude higher than the regular doses of radiotherapy. Therefore, the radiation dose should be well defined and precisely delivered to the target while minimizing radiation exposure to surrounding normal tissues [1]. Several methods have been proposed to obtain three-dimensional (3-D) dose distribution [2, 3]. In this paper, we propose an alternative method, which can be easily implemented in any stereotactic radiosurgery center with a magnetic resonance imaging (MRI) facility. A phantom with or without scattering centers filled with Fricke gel solution is irradiated with Gamma Knife(A (R)) system at a chosen spot. The phantom can be a replica of a human organ such as head, breast or any other organ. It can even be constructed from a real 3-D MR image of an organ of a patient using a computer-aided construction and irradiated at a specific region corresponding to the tumor position determined by MRI. The spin-lattice relaxation time T (1) of different parts of the irradiated phantom is determined by localized spectroscopy. The T (1)-weighted phantom images are used to correlate the image pixels intensity to the absorbed dose and consequently a 3-D dose distribution with a high resolution is obtained.