Hyperspectral Image Processing Algorithms Research Articles

NIR Spectral Imaging for Animal Feed Quality and Safety

Highlights Hyperspectral imaging can be a tool to address animal feed safety and quality issues. Various papers on hyperspectral imaging have emerged for grain and animal feed applications. More studies need to be done to use this technology in the field and in commercial scale. Abstract. The increasing need for safe and quality animal-derived food products and the limitations of slow and destructive traditional screening methods have driven the development of fast, accurate, and reliable testing methods for evaluation of animal feed safety and quality parameters. One of the methods being applied recently is near-infrared hyperspectral imaging (NIR HSI), an analytical technique that provides chemical and spatial information for the sample. It has been applied to various food and agricultural products and is gaining acceptance for commercial use in the food industry. This paper presents research progress with respect to applications of NIR HSI for safety and quality evaluation of animal feed ingredients, specifically in screening illegal adulteration and contamination of feed materials, detecting biological contamination and damage, measuring constituent concentrations and other quality properties, and evaluating feed processing parameters. Major results, multivariate analysis and hyperspectral image processing algorithms used, and limitations of the different studies being reviewed were highlighted. With its potential uses and applications on animal feed, NIR HSI is a promising breakthrough in the feed industry. Keywords: Animal feed, Feed safety, Feed quality, Hyperspectral imaging, NIR spectroscopy.

Geometrical Approximated Principal Component Analysis for Hyperspectral Image Analysis

Principal Component Analysis (PCA) is a method based on statistics and linear algebra techniques, used in hyperspectral satellite imagery for data dimensionality reduction required in order to speed up and increase the performance of subsequent hyperspectral image processing algorithms. This paper introduces the PCA approximation method based on a geometric construction approach (gaPCA) method, an alternative algorithm for computing the principal components based on a geometrical constructed approximation of the standard PCA and presents its application to remote sensing hyperspectral images. gaPCA has the potential of yielding better land classification results by preserving a higher degree of information related to the smaller objects of the scene (or to the rare spectral objects) than the standard PCA, being focused not on maximizing the variance of the data, but the range. The paper validates gaPCA on four distinct datasets and performs comparative evaluations and metrics with the standard PCA method. A comparative land classification benchmark of gaPCA and the standard PCA using statistical-based tools is also described. The results show gaPCA is an effective dimensionality-reduction tool, with performance similar to, and in several cases, even higher than standard PCA on specific image classification tasks. gaPCA was shown to be more suitable for hyperspectral images with small structures or objects that need to be detected or where preponderantly spectral classes or spectrally similar classes are present.

A Simulation Environment for Validation and Verification of Real Time Hyperspectral Processing Algorithms on-Board a UAV

The utilization of hyperspectral imaging sensors has gained a significant relevance among many different applications due to their capability for collecting a huge amount of information across the electromagnetic spectrum. These sensors have been traditionally mounted on-board satellites and airplanes in order to extract information from the Earth’s surface. Fortunately, the progressive miniaturization of these sensors during the last lustrum has enabled their use in other remote sensing platforms, such as drones equipped with hyperspectral cameras which bring advantages in terms of higher spatial resolution of the acquired images, more flexible revisit times and lower cost of the flight campaigns. However, when these drones are autonomously flying and taking real-time critical decisions from the information contained in the captured images, it is crucial that the whole process takes place in a safe and predictable manner. In order to deal with this problem, a simulation environment is presented in this work to analyze the virtual behavior of a drone equipped with a pushbroom hyperspectral camera used for assisting harvesting applications, which enables an exhaustive and realistic validation and verification of the drone real-time hyperspectral imaging system prior to its launch. To the best of the authors’ knowledge, the proposed environment represents the only solution in the state-of-the-art that allows the virtual verification of real-time hyperspectral image processing algorithms under realistic conditions.

Estimation of signal-dependent and -independent noise from hyperspectral images using a wavelet-based superpixel model

ABSTRACTNoise estimation is crucial for many hyperspectral (HS) image processing algorithms. In real HS images, the random noise is mainly composed of a signal-dependent (SD) photon noise component and a signal-independent (SI) electronic noise component. Based on a parametric model that accounts for the dependence of noise variance on the useful image signal, a novel method is proposed to estimate SD and SI noise variances in this paper. In order to accurately detect the homogeneous regions in noisy images, a new wavelet-based superpixel model is designed to segment a HS images into small patches that adhere to the local textures and hence persist in homogeneous characteristic. Then, the relevance vector machine (RVM) is exploited to split the noise and useful image signal in homogeneous superpixels. Finally, the SD and SI noise variances are obtained by fitting the scatter points of local means versus local total noise variances. Experiments on synthetic and real airborne visible/infrared imaging spectrometer (AVIRIS) HS images demonstrate the effectiveness of the proposed method.

Modified Residual Method for the Estimation of Noise in Hyperspectral Images

Many hyperspectral image processing algorithms (e.g., detection, classification, endmember extraction, and so on) are generally designed with the assumption of no spectral or spatial correlation in noise. However, previous studies have shown the presence of nonnegligible correlation between the noise samples in different spectral bands, especially between noises in adjacent bands, and that most of the well-known intrinsic dimension estimation algorithms give poor estimates in the presence of correlated noise. Thus, there is a need to tackle the specific case of spectrally correlated noise for noise estimation. We show, in this paper, that the commonly employed hyperspectral noise estimation algorithm based on regression residuals can be significantly affected by spectrally correlated noise and we suggest a modified approach that proves to be robust to noise correlation. Furthermore, the proposed method improves the noise variance estimates in comparison to the classic residual method even for the case of uncorrelated noise. Simulation results show that the estimation error is reduced at times by a factor of 5 when there is high spectral correlation in the noise. Our proposed per-pixel noise estimator requires an estimate of the noise covariance matrix, and for this, we also propose a method to estimate the noise covariance matrix. Simulation results demonstrate that the per-pixel noise estimates obtained via the use of estimated noise statistics are almost as good as those obtained via use of the true statistics.

Introduction to the Special Issue on High Performance Computing Solutions for Complex Problems

In the last decades, the complexity of the current and upcoming scientific/engineering problems has increased considerably. Computations involved in numerical simulations, molecular dynamics, computational fluid dynamics, bio-informatics, image processing, linear algebra, deep-learning, information retrieval, or big-data computing are just a few examples of such problems. At the same time, improvements in high performance computing (HPC) systems are mainly associated with an important increasing in the complexity of computer architectures, making difficult the code optimization. This results in a greater gap between the general scientific / engineering user community (in need of easy access to efficient high performance computations) and the HPC programmers community (who design codes for narrow sub-classes of problems). The development of user-friendly codes for non-HPC-trained user community becomes a big challenge. As consequence, effective use of HPC centers requires specialized / individual training for each user group. Today, programmers of HPC or/and scientific applications have to deal with numerous details regarding computer architectures at low level to take advantage of the last features of the current and upcoming computing systems. It makes difficult the efficient and optimized software development. Thus, strategies and tools that can help us to adapt our codes over different computing architectures is of vital importance. However, previously, we must know and identify what are most efficient programming strategies and architectonic features. This is a difficult task as both, strategies and features, depend on the particular problem to be dealt. Portability is other important issue today. Multiples kind of processors have arisen in the last years. Most of these current platforms use their own compilers, languages, etc., being a very complex task to implement portable codes. This special issue provides several studies, which involve different applications and strategies to improve performance and to achieve better usage of current computing systems. It is composed by three works. The work Performance Optimizations for an Automatic Target Generation Process in Hyperspectral Analysis by Fernando Sierra-Pajuelo, Abel Paz-Gallardo, and Antonio Plaza presents several optimizations for hyperspectral image processing algorithms intended to detect targets in hyperspectral images. The algorithm used is the automated target generation process (ATGP) and the optimizations comprise parallel versions of the algorithm developed using open multi-processing (OpenMP, including Intel Xeon Phi) and message passing interface (MPI). The study Using Computational Geometry to Improve Process Rescheduling on Round-Based Parallel Applications by R. da Rosa-Righi, V. Magalhaes-Guerreiro, G. Rostirolla, V. Facco-Rodrigues, C. Andre da Costa, and L. Dagnino-Chiwiacowsky, proposes two novel heuristics applied to process rescheduling, named MigCube and MigHull, to choose the candidate processes for migration and their destination. Both heuristics consider the use of computational geometry for plotting computation, communication and migration costs metrics in a 3D graph without any user intervention. Finally, the work Many-Task Computing on Many-Core Architectures by Pedro Valero-Lara, Poorima Nookala, Fernando L. Pelayo, Johan Jansoon, Serapheim Dimitropoulos, and Ioan Raicu, studies what are the Many-Task Computing (MTC) programming mechanisms to take advantages of the massively parallel features of current hardware accelerators for the particular target of MTC. Also, the hardware features of the two dominant many-core platforms (NVIDIA GPUs and Intel Xeon Phi) are also analyzed for our specific framework. This study consisted of comparing the time consumed for computing in parallel several tasks one by one (the whole computational resources are used just to compute one task at a time) with the time consumed for computing in parallel the same set of tasks simultaneously (the whole computational resources are used for computing the set of tasks at very same time). Finally, both software-hardware scenarios were compared to identify the most relevant computer features in each of our many-core architectures. We would like to thank the editorial board of SCPE and reviewers for their effort and time, which is very appreciated.

Performance Optimizations for an Automatic Target Generation Process in Hyperspectral Analysis

Hyperspectral sensors acquire images with hundreds of spectral channels. These images have a lot of information in both spectral and spatial domain, and with this kind of information different research studies can be accomplished. In this work, we present several optimizations for hyperspectral image processing algorithms intended to detect targets in hyperspectral images. The hyperspectral image selected for our study was collected by the NASAs Airborne Visible Infra-Red Imaging Spectrometer (AVIRIS) over the World Trade Center (WTC) in New York, five days after September 11th attack. The algorithm used in our experiments is the automated target generation process (ATGP) and our optimizations comprise parallel versions of the algorithm developed using open multi-processing (OpenMP) and message passing interface (MPI). Our experiments indicate that the ATGP can be successfully implemented in parallel in multicore and cluster computing architectures, including Intel Xeon Phi.

Combining spectral and spatial information from aerial hyperspectral images for delineating homogenous management zones

Most available hyperspectral image processing algorithms analyse the data based on the spectral information exclusively and do not treat the data as an image. Methods that use spatial and spectral information mostly perform a two-step processing technique, i.e. the spatial analysis is used as a retrospective smoothing. In this paper a recently developed method, which is based on the beamlet analysis, is presented. This method, which uses spectral and spatial information simultaneously, was adjusted in this study to multi- and hyperspectral images. The results of the new methodology, called the recursive dyadic partitioning=beamlet decorated (RDP-BD), were compared to fuzzy C-means spectral classification and to spatially coherent regions classification. These methods were used to classify nitrogen levels in an experimental potato plot, which was grown under different nitrogen treatments. The image was classified by these three methods based on the combination of two spectral indexes: transformed chlorophyll absorption reflectance index (TCARI) and optimized soil-adjust vegetation index (OSAVI), the first three principal components of the image and the complete 210 bands of the hyperspectral image. Although the spatial analysis slightly improved the classification (from 71% to 78%), the most pronounced contribution was its ability to divide the experimental plot into zones that fit the treatments borders. In addition, it was shown that the principle components and the whole spectra out-performed the spectral index. These methods should assist with testing the efficiency of variable rate nitrogen fertilisation as they utilise the informative spectral data to properly classify nitrogen level in the pixel scale and delineate management zones at the field scale.

FPGA Implementation of Abundance Estimation for Spectral Unmixing of Hyperspectral Data Using the Image Space Reconstruction Algorithm

One of the most popular and widely used techniques for analyzing remotely sensed hyperspectral data is spectral unmixing, which relies on two stages: (i) identification of pure spectral signatures (endmembers) in the data, and (ii) estimation of the abundance of each endmember in each (possibly mixed) pixel. Due to the high dimensionality of the hyperspectral data, spectral unmixing is a very time-consuming task. With recent advances in reconfigurable computing, especially using field programmable gate arrays (FPGAs), hyperspectral image processing algorithms can now be accelerated for on-board exploitation using compact hardware components with small size and cost. Although in previous work several efforts have been directed towards FPGA implementation of endmember extraction algorithms, the abundance estimation step has received comparatively much less attention. In this work, we develop a parallel FPGA-based design of the image space reconstruction algorithm (ISRA), a technique for solving linear inverse problems with positive constraints that has been used to estimate the abundance of each endmember in each pixel of a hyperspectral image. It is an iterative algorithm that guarantees convergence (after a certain number of iterations) and positive values in the results of the abundances (an important consideration in unmixing applications). Our system includes a direct memory access (DMA) module and implements a pre-fetching technique to hide the latency of the input/output communications. The method has been implemented on a Virtex-4 XC4VFX60 FPGA (a model that is similar to radiation-hardened FPGAs certified for space operation) and tested using real hyperspectral data sets collected by the Airborne Visible Infra-Red Imaging Spectrometer (AVIRIS) over the Cuprite mining district in Nevada and the Jasper Ridge Biological Preserve in California. Experimental results demonstrate that our hardware version can significantly outperform an equivalent software version, thus being able to provide abundance estimation results in near real-time, which makes our reconfigurable system appealing for on-board hyperspectral data processing.

Pyrite mine waste and water mapping using Hymap and Hyperion hyperspectral data

Hyperspectral low spatial resolution Hyperion data are used to map mine waste from massive sulphide ore deposits, mostly abandoned, on the Iberian Pyrite Belt. Hymap high spatial resolution data are used for crossed interpretation. Mine dams, mill tailings and mine dumps in a variable state of pyrite oxidation are recognizable. Ponds of acid water in the mine sites are spectrally outstanding and, with simple image processing procedures, mappable. In addition, acid water with different chemical composition is mappable with hyperspectral data, whether of high or low spatial resolution. A sequence of hyperspectral image processing algorithms used to produce the maps is suggested. Therefore, hyperspectral data are invaluable in giving quick hints on the quality of the rapidly changing state of the contamination generated by sulphide mine waste, enabling the authorities to activate mitigation procedures.

Improving the Performance of Hyperspectral Image and Signal Processing Algorithms Using Parallel, Distributed and Specialized Hardware-Based Systems

Advances in sensor technology are revolutionizing the way remotely sensed data is collected, managed and analyzed. The incorporation of latest-generation sensors to airborne and satellite platforms is currently producing a nearly continual stream of high-dimensional data, and this explosion in the amount of collected information has rapidly created new processing challenges. For instance, hyperspectral signal processing is a new technique in remote sensing that generates hundreds of spectral bands at different wavelength channels for the same area on the surface of the Earth. Many current and future applications of remote sensing in Earth science, space science, and soon in exploration science will require (near) real-time processing capabilities. In recent years, several efforts have been directed towards the incorporation of high-performance computing (HPC) systems and architectures in remote sensing missions. With the aim of providing an overview of current and new trends in parallel and distributed systems for remote sensing applications, this paper explores three HPC-based paradigms for efficient implementation of the Pixel Purity Index (PPI) algorithm, available from the popular Kodak's Research Systems ENVI software package, as a representative case study for demonstration purposes. Several different parallel programming techniques are used to improve the performance of the PPI on a variety of parallel platforms, including a set of message passing interface (MPI)-based implementations on a massively parallel Beowulf cluster at NASA's Goddard Space Flight Center in Maryland and on a variety of heterogeneous networks of workstations at University of Maryland; a Handel-C implementation of the algorithm on a Virtex-II field programmable gate array (FPGA); and a compute unified device architecture (CUDA)-based implementation on graphical processing units (GPUs) of NVidia. Combined, these parts deliver an excellent snapshot of the state-of-the-art in those areas, and offer a thoughtful perspective on the potential and emerging challenges of adapting HPC systems to remote sensing problems.

HYPERSPECTRAL IMAGING FOR DETECTING FECAL AND INGESTA CONTAMINANTS ON POULTRY CARCASSES

A hyperspectral imaging system including a camera with prismgratingprism spectrograph, fiber optic linelighting, motorized lens control, and hyperspectral imageprocessing software was developed for poultry safety inspection,particularly identification of fecal and ingesta contamination on poultry carcasses. Both spectral and spatial image databetween 400 and 900 nm with 512 spectral bands were acquired from fecal and ingesta contaminated poultry carcasses. Fourdominant wavelengths (434, 517, 565, and 628 nm) were selected by principal component analysis from visible/nearinfraredspectroscopy for wavelength selection of hyperspectral images. A calibration model for the hyperspectral imaging systemwas developed from calibration lighting sources (HgAr, Kr, and lasers) for accurate band selection from the hyperspectralimages to identify spatial and spectral characterization of fecal and ingesta contaminants. Hyperspectral image processingalgorithms, specifically band ratio of dualwavelength (565/517) images and histogram stretching, were effective in theidentification of fecal and ingesta contamination of poultry carcasses. Test results indicated that the detection accuracy was97.3% for linear and 100% for nonlinear histogram stretching. This article presents the research results that hyperspectralimaging can be used effectively for detecting feces (from the duodenum, ceca, and colon) and ingesta on poultry carcassesand demonstrates the potential application for online safety inspection of poultry.