Applying ISOMAP to the learning of hyperspectral image

In this paper, we present the application of a non-linear dimensionality reduction technique for the learning and probabilistic classification of hyperspectral image. Hyperspectral image spectroscopy is an emerging technique for geological investigations from airborne or orbital sensors. It gives much greater information content per pixel on the image than a normal colour image. This should greatly help with the autonomous identification of natural and manmade objects in unfamiliar terrains for robotic vehicles. However, the large information content of such data makes interpretation of hyperspectral images time-consuming and userintensive. We propose the use of Isomap, a non-linear manifold learning technique combined with Expectation Maximisation in graphical probabilistic models for learning and classification. Isomap is used to find the underlying manifold of the training data. This low dimensional representation of the hyperspectral data facilitates the learning of a Gaussian Mixture Model representation, whose joint probability distributions can be calculated offline. The learnt model is then applied to the hyperspectral image at runtime and data classification can be performed.

Hyperspectral image compression onboard next-generation satellites: implementation solutions on GPU and FPGAs

Programa de doctorado: Ingeniería de Telecomunicación Avanzada

Airborne hyperspectral image data of Heron Reef, Australia

This airborne hyperspectral (19 bands) image data of Heron Reef, Great Barrier Reef, Australia is derived from Compact Airborne Spectrographic Imager (CASI) data acquired on 1st and 3rd of July 2002, latitude -23.45, longitude 151.92. Processing and correction to at-surface data was completed by Karen Joyce (Joyce, 2004). Raw imagery consisted several images corresponding to the number of flight paths taken to cover the entire Heron Reef. Spatial resolution is one meter. Radiometric corrections converted the at-sensor digital number values to at surface spectral radiance values using sensor specific calibration coefficients and CSIRO's c-WomBat-c atmospheric correction software. Geometric corrections were done using field collected coordinates of features identified in the image. Projection used was Universal Transverse Mercator Zone 56 South and Datum used was WGS 84. Image data is in TIFF format.

Monitoring spinach shelf-life with hyperspectral image through

Different procedures for monitoring the evolution of leafy vegetables, under plastic covers during cold storage, have been studied. Fifteen spinach leaves were put inside Petri dishes covered with three different plastic films and stored at 4 °C for 21 days. Hyperspectral images were taken during this storage. A radiometric correction is proposed in order to avoid the variation in transmittance of the plastic films during time in the hyperspectral images. Afterwards, three spectral pre-processing procedures (no pre-process, Savitsky–Golay and Standard Normal Variate, combined with Principal Component Analysis) were applied to obtain different models. The corresponding artificial images of scores were studied by means of Analysis of Variance to compare their ability to sense the aging of the leaves. All models were able to monitor the aging through storage. Radiometric correction seemed to work properly and could allow the supervision of shelf-life in leafy vegetables through commercial transparent films.

Application of Improved Direct Calibration for Hyperspectral image processing: Detecting peanut traces in wheat flour.

NIR Hyperspectral imaging (1000-2500 nm) combined with IDC allowed the detection of peanut traces down to adulteration percentages 0.01% Contrary to PLSR, IDC does not require a calibration set, but uses both expert and experimental information and suitable for quantification of an interest compound in complex matrices. The obtained results shows the feasibility of using HSI systems for the detection of peanut traces in conjunction with chemical procedures, such as RT-PCR and ELISA

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.

Applying incremental EM to Bayesian classifiers in the learning of hyperspectral remote sensing data

In this paper, we apply the incremental EM method to Bayesian Network Classifiers to learn and interpret hyperspectral sensor data in robotic planetary missions. Hyperspectral image spectroscopy is an emerging technique for geological investigations from airborne or orbital sensors. Many spacecraft carry spectroscopic equipment as wavelengths outside the visible light in the electromagnetic spectrum give much greater information about an object. The algorithm used is an extension to the standard Expectation Maximisation (EM). The incremental method allows us to learn and interpret the data as they become available. Two Bayesian network classifiers were tested: the Naive Bayes, and the Tree-Augmented-Naive Bayes structures. Our preliminary experiments show that incremental learning with unlabelled data can improve the accuracy of the classifier.

Maize kernel hardness classification by near infrared (NIR) hyperspectral imaging and multivariate data analysis.

The use of near infrared (NIR) hyperspectral imaging and hyperspectral image analysis for distinguishing between hard, intermediate and soft maize kernels from inbred lines was evaluated. NIR hyperspectral images of two sets (12 and 24 kernels) of whole maize kernels were acquired using a Spectral Dimensions MatrixNIR camera with a spectral range of 960-1662 nm and a sisuChema SWIR (short wave infrared) hyperspectral pushbroom imaging system with a spectral range of 1000-2498 nm. Exploratory principal component analysis (PCA) was used on absorbance images to remove background, bad pixels and shading. On the cleaned images. PCA could be used effectively to find histological classes including glassy (hard) and floury (soft) endosperm. PCA illustrated a distinct difference between glassy and floury endosperm along principal component (PC) three on the MatrixNIR and PC two on the sisuChema with two distinguishable clusters. Subsequently partial least squares discriminant analysis (PLS-DA) was applied to build a classification model. The PLS-DA model from the MatrixNIR image (12 kernels) resulted in root mean square error of prediction (RMSEP) value of 0.18. This was repeated on the MatrixNIR image of the 24 kernels which resulted in RMSEP of 0.18. The sisuChema image yielded RMSEP value of 0.29. The reproducible results obtained with the different data sets indicate that the method proposed in this paper has a real potential for future classification uses.

Contributions to the analysis and segmentation of remote sensing hyperspectral images

142 p.

System-on-chip field-programmable gate array design for onboard real-time hyperspectral unmixing

Hyperspectral instruments have been incorporated in satellite missions, providing large amounts of data of high spectral resolution of the Earth surface. This data can be used in remote sensing applications that often require a real-time or near-real-time response. To avoid delays between hyperspectral image acquisition and its interpretation, the last usually done on a ground station, onboard systems have emerged to process data, reducing the volume of information to transfer from the satellite to the ground station. For this purpose, compact reconfigurable hardware modules, such as field-programmable gate arrays (FPGAs), are widely used. This paper proposes an FPGA-based architecture for hyperspectral unmixing. This method based on the vertex component analysis (VCA) and it works without a dimensionality reduction preprocessing step. The architecture has been designed for a low-cost Xilinx Zynq board with a Zynq-7020 system-on-chip FPGA-based on the Artix-7 FPGA programmable logic and tested using real hyperspectral data. Experimental results indicate that the proposed implementation can achieve real-time processing, while maintaining the methods accuracy, which indicate the potential of the proposed platform to implement high-performance, low-cost embedded systems, opening perspectives for onboard hyperspectral image processing.

Urban Detection From Hyperspectral Images Using Dimension-Reduction Model and Fusion of Multiple Segmentations Based on Stuctural and Textural Features

Ce mémoire de maîtrise présente une nouvelle approche non supervisée pour détecter et segmenter les régions urbaines dans les images hyperspectrales. La méthode proposée n ́ecessite trois étapes. Tout d’abord, afin de réduire le coût calculatoire de notre algorithme, une image couleur du contenu spectral est estimée. A cette fin, une étape de réduction de dimensionalité non-linéaire, basée sur deux critères complémentaires mais contradictoires de bonne visualisation; à savoir la précision et le contraste, est réalisée pour l’affichage couleur de chaque image hyperspectrale. Ensuite, pour discriminer les régions urbaines des régions non urbaines, la seconde étape consiste à extraire quelques caractéristiques discriminantes (et complémentaires) sur cette image hyperspectrale couleur. A cette fin, nous avons extrait une série de paramètres discriminants pour décrire les caractéristiques d’une zone urbaine, principalement composée d’objets manufacturés de formes simples g ́eométriques et régulières. Nous avons utilisé des caractéristiques texturales basées sur les niveaux de gris, la magnitude du gradient ou des paramètres issus de la matrice de co-occurrence combinés avec des caractéristiques structurelles basées sur l’orientation locale du gradient de l’image et la détection locale de segments de droites. Afin de réduire encore la complexité de calcul de notre approche et éviter le problème de la ”malédiction de la dimensionnalité” quand on décide de regrouper des données de dimensions élevées, nous avons décidé de classifier individuellement, dans la dernière étape, chaque caractéristique texturale ou structurelle avec une simple procédure de K-moyennes et ensuite de combiner ces segmentations grossières, obtenues à faible coût, avec un modèle efficace de fusion de cartes de segmentations. Les expérimentations données dans ce rapport montrent que cette stratégie est efficace visuellement et se compare favorablement aux autres méthodes de détection et segmentation de zones urbaines à partir d’images hyperspectrales.

FPGA-based architecture for hyperspectral endmember extraction

Hyperspectral instruments have been incorporated in satellite missions, providing data of high spectral resolution of the Earth. This data can be used in remote sensing applications, such as, target detection, hazard prevention, and monitoring oil spills, among others. In most of these applications, one of the requirements of paramount importance is the ability to give real-time or near real-time response. Recently, onboard processing systems have emerged, in order to overcome the huge amount of data to transfer from the satellite to the ground station, and thus, avoiding delays between hyperspectral image acquisition and its interpretation. For this purpose, compact reconfigurable hardware modules, such as field programmable gate arrays (FPGAs) are widely used. This paper proposes a parallel FPGA-based architecture for endmember’s signature extraction. This method based on the Vertex Component Analysis (VCA) has several advantages, namely it is unsupervised, fully automatic, and it works without dimensionality reduction (DR) pre-processing step. The architecture has been designed for a low cost Xilinx Zynq board with a Zynq-7020 SoC FPGA based on the Artix-7 FPGA programmable logic and tested using real hyperspectral data sets collected by the NASA’s Airborne Visible Infra-Red Imaging Spectrometer (AVIRIS) over the Cuprite mining district in Nevada. Experimental results indicate that the proposed implementation can achieve real-time processing, while maintaining the methods accuracy, which indicate the potential of the proposed platform to implement high-performance, low cost embedded systems, opening new perspectives for onboard hyperspectral image processing.

Statistical analysis of spectral data: A methodology for designing an intelligent monitoring system for the diabetic foot

Early detection of (pre-)signs of ulceration on a diabetic foot is valuable for clinical practice. Hyperspectral imaging is a promising technique for detection and classification of such (pre-)signs. However, the number of the spectral bands should be limited to avoid overfitting, which is critical for pixel classification with hyperspectral image data. The goal was to design a detector/classifier based on spectral imaging (SI) with a small number of optical bandpass filters. The performance and stability of the design were also investigated. The selection of the bandpass filters boils down to a feature selection problem. A dataset was built, containing reflectance spectra of 227 skin spots from 64 patients, measured with a spectrometer. Each skin spot was annotated manually by clinicians as "healthy" or a specific (pre-)sign of ulceration. Statistical analysis on the data set showed the number of required filters is between 3 and 7, depending on additional constraints on the filter set. The stability analysis revealed that shot noise was the most critical factor affecting the classification performance. It indicated that this impact could be avoided in future SI systems with a camera sensor whose saturation level is higher than 106, or by postimage processing.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and deterministic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel metaheuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS metaheuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.