GPU implementation of RX detection using spectral derivative features

Hyperspectral image (HSI), which can record abundance information of a pixel, has shown huge potential on many applications such as image classification, target and anomaly detection and so on. Nowadays, anomaly detection has attracted more attention because there is no limitation of spectral library. A standard approach for anomaly detection is the method developed by Reed and Xiaoli, called RX algorithm. However, the data volume is getting bigger with the developing of imaging technology. A problem that ensues is the rapid increase of computation complexity and this will lead a time-consumed application. In addition, there will be noise in HSI with the influence of illumination and atmospheric. In this paper, we realize an implementation of RX algorithm on NVIDIA GeForce 1060 GPU with the utilization of derivative features. On one hand, the GPU parallel implementation reach the purpose of real-time processing and it also eliminates the storage burden of on-board processing. On the other hand, the derivative features have better performance on salient features detection and noise restraint. Thus, it can further promote the detection performance of RXD. In our experiments, three real HSI datasets were tested to verify the effect of GPU parallel implementation. The experiment results had indicated that the utilization of derivative features can promote the detection performance. Compared with serial computation, the parallel implementation achieves a great reduction on processing time.     

Rapid real-time generation of super-resolution hyperspectral images through compressive sensing and GPU

Recently, compressive sensing (CS) has offered a new framework whereby a signal can be recovered from a small number of noisy non-adaptive samples. This is now an active area of research in many image-processing applications, especially super-resolution. CS algorithms are widely known to be computationally expensive. This paper studies a real time super-resolution reconstruction method based on the compressive sampling matching pursuit (CoSaMP) algorithm for hyperspectral images. CoSaMP is an iterative compressive sensing method based on the orthogonal matching pursuit (OMP). Multi-spectral images record enormous volumes of data that are required in practical modern remote-sensing applications. A proposed implementation based on the graphical processing unit (GPU) has been developed for CoSaMP using computed unified device architecture (CUDA) and the cuBLAS library. The CoSaMP algorithm is divided into interdependent parts with respect to complexity and potential for parallelization. The proposed implementation is evaluated in terms of reconstruction error for different state-of-the-art super-resolution methods. Various experiments were conducted using real hyperspectral images collected by Earth Observing-1 (EO-1), and experimental results demonstrate the speeding up of the proposed GPU implementation and compare it to the sequential CPU implementation and state-of-the-art techniques. The speeding up of the GPU-based implementation is up to approximately 70 times faster than the corresponding optimized CPU.     

FPGA implementation of collaborative representation algorithm for real-time hyperspectral target detection

Hyperspectral image contains various wavelength channels and the corresponding imagery processing requires a computation platform with high performance. Target and anomaly detection on hyperspectral image has been concerned because of its practicality in many real-time detection fields while wider applicability is limited by the computing condition and low processing speed. The field programmable gate arrays (FPGAs) offer the possibility of on-board hyperspectral data processing with high speed, low-power consumption, reconfigurability and radiation tolerance. In this paper, we develop a novel FPGA-based technique for efficient real-time target detection algorithm in hyperspectral images. The collaborative representation is an efficient target detection (CRD) algorithm in hyperspectral imagery, which is directly based on the concept that the target pixels can be approximately represented by its spectral signatures, while the other cannot. To achieve high processing speed on FPGAs platform, the CRD algorithm reduces the dimensionality of hyperspectral image first. The Sherman–Morrison formula is utilized to calculate the matrix inversion to reduce the complexity of overall CRD algorithm. The achieved results demonstrate that the proposed system may obtains shorter processing time of the CRD algorithm than that on 3.40 GHz CPU.     

RX architectures for real-time anomaly detection in hyperspectral images

In the field of hyperspectral image processing, anomaly detection (AD) is a deeply investigated task whose goal is to find objects in the image that are anomalous with respect to the background. In many operational scenarios, detection, classification and identification of anomalous spectral pixels have to be performed in real time to quickly furnish information for decision-making. In this framework, many studies concern the design of computationally efficient AD algorithms for hyperspectral images in order to assure real-time or nearly real-time processing. In this work, a sub-class of anomaly detection algorithms is considered, i.e., those algorithms aimed at detecting small rare objects that are anomalous with respect to their local background. Among such techniques, one of the most established is the Reed–Xiaoli (RX) algorithm, which is based on a local Gaussian assumption for background clutter and locally estimates its parameters by means of the pixels inside a window around the pixel under test (PUT). In the literature, the RX decision rule has been employed to develop computationally efficient algorithms tested in real-time systems. Initially, a recursive block-based parameter estimation procedure was adopted that makes the RX processing and the detection performance differ from those of the original RX. More recently, an update strategy has been proposed which relies on a line-by-line processing without altering the RX detection statistic. In this work, the above-mentioned RX real-time oriented techniques have been improved using a linear algebra-based strategy to efficiently update the inverse covariance matrix thus avoiding its computation and inversion for each pixel of the hyperspectral image. The proposed strategy has been deeply discussed pointing out the benefits introduced on the two analyzed architectures in terms of overall number of elementary operations required. The results show the benefits of the new strategy with respect to the original architectures.     

Fast anomaly detection in hyperspectral images with RX method on heterogeneous clusters

Remotely sensed hyperspectral sensors provide image data containing rich information in both the spatial and the spectral domain, and this information can be used to address detection tasks in many applications. One of the most widely used and successful algorithms for anomaly detection in hyperspectral images is the RX algorithm. Despite its wide acceptance and high computational complexity when applied to real hyperspectral scenes, few approaches have been developed for parallel implementation of this algorithm. In this paper, we evaluate the suitability of using a hybrid parallel implementation with a high-dimensional hyperspectral scene. A general strategy to automatically map parallel hybrid anomaly detection algorithms for hyperspectral image analysis has been developed. Parallel RX has been tested on an heterogeneous cluster using this routine. The considered approach is quantitatively evaluated using hyperspectral data collected by the NASA’s Airborne Visible Infra-Red Imaging Spectrometer system over the World Trade Center in New York, 5 days after the terrorist attacks. The numerical effectiveness of the algorithms is evaluated by means of their capacity to automatically detect the thermal hot spot of fires (anomalies). The speedups achieved show that a cluster of multi-core nodes can highly accelerate the RX algorithm.     

A novel FPGA-based architecture for the estimation of the virtual dimensionality in remotely sensed hyperspectral images

A challenging problem in spectral unmixing is how to determine the number of endmembers in a given scene. One of the most popular ways to determine the number of endmembers is by estimating the virtual dimensionality (VD) of the hyperspectral image using the well-known Harsanyi–Farrand–Chang (HFC) method. Due to the complexity and high dimensionality of hyperspectral scenes, this task is computationally expensive. Reconfigurable field-programmable gate arrays (FPGAs) are promising platforms that allow hardware/software codesign and the potential to provide powerful onboard computing capabilities and flexibility at the same time. In this paper, we present the first FPGA design for the HFC-VD algorithm. The proposed method has been implemented on a Virtex-7 XC7VX690T FPGA and tested using real hyperspectral data collected by NASA’s Airborne Visible Infra-Red Imaging Spectrometer over the Cuprite mining district in Nevada and the World Trade Center in New York. Experimental results demonstrate that our hardware version of the HFC-VD algorithm can significantly outperform an equivalent software version, which makes our reconfigurable system appealing for onboard hyperspectral data processing. Most important, our implementation exhibits real-time performance with regard to the time that the hyperspectral instrument takes to collect the image data.     

Parallel probabilistic relaxation labelling based on Markov random fields for spectral-spatial hyperspectral image classification

The large volume of data and computational complexity of algorithms limit the application of hyperspectral image classification to real-time operations. This work addresses the use of different parallel processing techniques to speed up the Markov random field (MRF)-based method to perform spectral-spatial classification of hyperspectral imagery. The Metropolis relaxation labelling approach is modified to take advantage of multi-core central processing units (CPUs) and to adapt it to massively parallel processing systems like graphics processing units (GPUs). The experiments on different hyperspectral data sets revealed that the implementation approach has a huge impact on the execution time of the algorithm. The results demonstrated that the modified MRF algorithm produced classification accuracy similar to conventional methods with greatly improved computational performance. With modern multi-core CPUs, good computational speed-up can be achieved even without additional hardware support. The CPU-GPU hybrid framework rendered the otherwise computationally expensive approach suitable for time-constrained applications.     

基于CPU/GPU异构模式的高光谱遥感影像数据处理研究与实现

近年来,基于GPU的新型异构高性能计算模式的蓬勃发展为众多领域应用提供了良好的发展机遇,国内外遥感专家开始引入高性能异构计算来解决高光谱遥感影像高维空间特点所带来的数据计算量大、实时处理难等问题。在此简要介绍了高光谱遥感和CPU/GPU异构计算模式,总结了近几年国内外基于CPU/GPU异构模式的高光谱遥感数据处理研究现状和问题;并面向共享存储型小型桌面超级计算机,基于CPU/GPU异构模式实现了高光谱遥感影像MNF降维的并行化,通过与串行程序和共享存储的OpenMP同构模式对比,验证了异构模式在高光谱遥感处理领域的发展潜力。     

一种利用均值匹配改进的高光谱异常检测方法

传统高光谱异常检测算法由于背景信息估计不准确等原因普遍存在高虚警率的问题,针对这一现象,提出了一种利用图像均值进行匹配改进的高光谱异常目标检测后验处理方法。首先采用传统的高光谱异常检测算法将待检测高光谱图像划分为背景与异常目标潜在区域,之后通过对待测图像求解均值,将其与异常目标潜在区域像元进行相似性匹配计算,剔除大范围误检像元,得到最终检测结果。该方法在传统异常目标检测算法基础上进行相似度量剔除大范围虚警像元,在提高原算法探测能力的同时有效地降低虚警率。实验表明,该方法可以有效降低虚警率,提高原算法对于亚像元异常目标的检测能力,且对于不同算法、不同数据具有普适性。     

基于高维几何特性的高光谱异常检测算法研究

提出了一种新的高光谱图像异常检测算法。作为一种多元数据集合,高光谱数据一般呈现出共超平面的几何特性,我们利用这一特点来求取垂直于超平面的法线矢量,并将数据投影到这一法线矢量方向,从而分离出异常点,达到异常检测的目的。本算法适合于对小目标的检测,且不需要先验的光谱信息。对算法的可行性进行了仿真并将它应用于高光谱数据,取得了较好的结果。     

结合孪生网络和像素配对的高光谱图像异常检测

目的 高光谱遥感中,通常利用像素的光谱特征来区分背景地物和异常目标,即通过二者之间的光谱差异来寻找图像中的异常像元。但传统的异常检测算法并未有效挖掘光谱的深层特征,高光谱图像中丰富的光谱信息没有被充分利用。针对这一问题,本文提出结合孪生神经网络和像素配对策略的高光谱图像异常检测方法,利用深度学习技术提取高光谱图像的深层非线性特征,提高异常检测精度。方法 采用像素配对的思想构建训练样本,与原始数据集相比,配对得到的新数据集数量呈指数增长,从而满足深度网络对数据集数量的需求。搭建含有特征提取模块和特征处理模块的孪生网络模型,其中,特征处理模块中的卷积层可以专注于提取像素对之间的差异特征,随后利用新的训练像素对数据集进行训练,并将训练好的分类模型固定参数,迁移至检测过程。用滑动双窗口策略对测试集进行配对处理,将测试像素对数据集送入网络模型,得到每个像素相较于周围背景像素的差异性分数,从而识别测试场景中的异常地物。结果 在异常检测的实验结果中,本文提出的孪生网络模型在San Diego数据集的两幅场景和ABU-Airport数据集的一幅场景上,得到的AUC （area under the curve）值分别为0.993 51、0.981 21和0.984 38,在3个测试集上的表现较传统方法和基于卷积神经网络的异常检测算法具有明显优势。结论 本文方法可以提取输入像素对的深层光谱特征,并根据其特征的差异性,让网络学习到二者的区分度,从而更好地赋予待测像素相对于周围背景的异常分数。本文方法相对于卷积神经网络的异常检测方法可以有效地降低虚警,与传统方法相比能够更加明显地突出异常目标,提高了检测率,同时也具有较强的鲁棒性。     

粗定位和协同表示的高光谱图像异常检测

目的 由于在军事和民用应用中的重要作用,高光谱遥感影像异常检测在过去的20~30年里一直都是备受关注的研究热点。然而,考虑到异常点往往藏匿于大量的背景像元之中,且只占据很少的数量,给精确检测带来了不小的挑战。针对此问题,基于异常点往往表现在高频的细节区域这一前提,本文提出了一种基于异常点粗定位和协同表示的高光谱遥感影像异常检测算法。方法 对输入的原始高光谱遥感影像进行空间维的降质操作;通过衡量降质后影像与原始影像在空间维的差异,粗略定位可能的异常点位置;将粗定位的异常点位置用于指导像元间的协同表示以重构像元;通过衡量重构像元与原始像元的差异,从而进一步优化异常检测结果。结果 在4个数据集上与6种方法进行了实验对比。对于San Diego数据集,次优算法和本文算法分别取得的AUC （area under curve）值为0.978 6和0.994 0;对于HYDICE （hyperspectral digital image collection equipment）数据集,次优算法和本文算法的AUC值为0.993 6和0.998 5;对于Honghu数据集,次优算法和本文方法的AUC值分别为0.999 2和0.999 3;对Grand Isle数据集而言,尽管本文方法以0.001的差距略低于性能第1的算法,但从目视结果图中可见,本文方法所产生的虚警目标远少于性能第1的算法。结论 本文所提出的粗定位和协同表示的高光谱异常检测算法,综合考虑了高光谱遥感影像的谱间特性,同时还利用了其空间特性以及空间信息的先验分布,从而获得异常检测结果的提升。     

Parallel techniques for information extraction from hyperspectral imagery using heterogeneous networks of workstations

Recent advances in space and computer technologies are revolutionizing the way remotely sensed data is collected, managed and interpreted. In particular, NASA is continuously gathering very high-dimensional imagery data from the surface of the Earth with hyperspectral sensors such as the Jet Propulsion Laboratory's airborne visible-infrared imaging spectrometer (AVIRIS) or the Hyperion imager aboard Earth Observing-1 (EO-1) satellite platform. The development of efficient techniques for extracting scientific understanding from the massive amount of collected data is critical for space-based Earth science and planetary exploration. In particular, many hyperspectral imaging applications demand real time or near real-time performance. Examples include homeland security/defense, environmental modeling and assessment, wild-land fire tracking, biological threat detection, and monitoring of oil spills and other types of chemical contamination. Only a few parallel processing strategies for hyperspectral imagery are currently available, and most of them assume homogeneity in the underlying computing platform. In turn, heterogeneous networks of workstations (NOWs) have rapidly become a very promising computing solution which is expected to play a major role in the design of high-performance systems for many on-going and planned remote sensing missions. In order to address the need for cost-effective parallel solutions in this fast growing and emerging research area, this paper develops several highly innovative parallel algorithms for unsupervised information extraction and mining from hyperspectral image data sets, which have been specifically designed to be run in heterogeneous NOWs. The considered approaches fall into three highly representative categories: clustering, classification and spectral mixture analysis. Analytical and experimental results are presented in the context of realistic applications (based on hyperspectral data sets from the AVIRIS data repository) using several homogeneous and heterogeneous parallel computing facilities available at NASA's Goddard Space Flight Center and the University of Maryland.     

Parallel unmixing of remotely sensed hyperspectral images on commodity graphics processing units

Hyperspectral imaging instruments are capable of collecting hundreds of images, corresponding to different wavelength channels, for the same area on the surface of the Earth. One of the main problems in the analysis of hyperspectral data cubes is the presence of mixed pixels, which arise when the spatial resolution of the sensor is not enough to separate spectrally distinct materials. Hyperspectral unmixing is one of the most popular techniques to analyze hyperspectral data. It comprises two stages: (i) automatic identification of pure spectral signatures (endmembers) and (ii) estimation of the fractional abundance of each endmember in each pixel. The spectral unmixing process is quite expensive in computational terms, mainly due to the extremely high dimensionality of hyperspectral data cubes. Although this process maps nicely to high performance systems such as clusters of computers, these systems are generally expensive and difficult to adapt to real‐time data processing requirements introduced by several applications, such as wildland fire tracking, biological threat detection, monitoring of oil spills, and other types of chemical contamination. In this paper, we develop an implementation of the full hyperspectral unmixing chain on commodity graphics processing units (GPUs). The proposed methodology has been implemented, using the CUDA (compute device unified architecture), and tested on three different GPU architectures: NVidia Tesla C1060, NVidia GeForce GTX 275, and NVidia GeForce 9800 GX2, achieving near real‐time unmixing performance in some configurations tested when analyzing two different hyperspectral images, collected over the World Trade Center complex in New York City and the Cuprite mining district in Nevada. Copyright © 2011 John Wiley & Sons, Ltd.     

Band Selection of Hyperspectral Images for Automatic Detection of Poultry Skin Tumors

This paper presents a spectral band selection method for feature dimensionality reduction in hyperspectral image analysis for detecting skin tumors on poultry carcasses. A hyperspectral image contains spatial information measured as a sequence of individual wavelength across broad spectral bands. Despite the useful information for skin tumor detection, real-time processing of hyperspectral images is often a challenging task due to the large amount of data. Band selection finds a subset of significant spectral bands in terms of information content for dimensionality reduction. This paper presents a band selection method of hyperspectral images based on the recursive divergence for the automatic detection of poultry carcasses. For this, we derive a set of recursive equations for the fast calculation of divergence with an additional band to overcome the computational restrictions in real-time processing. A support vector machine is used as a classifier for tumor detection. From our experiments, the proposed band selection method shows high detection accuracy with low false positive rates compared to the canonical analysis at a small number of spectral bands. Also, compared with the enumeration approach of 93.75% detection rate, our proposed recursive divergence approach gives 90.6% detection rate, which is within the industry-accepted accuracy of 90-95%, while achieving the computational saving for real-time processing.     

高光谱遥感蚀变填图SCM并行算法设计与实现

由于空间和波谱分辨率的不断提高,高光谱遥感影像的海量数据特性导致高光谱遥感影像并行处理成为遥感影像处理技术的发展趋势。本文基于CUDA和GPU环境,设计并实现了高光谱遥感蚀变填图的SCM并行算法。实验结果表明,并行加速比可达到25,SCM并行算法能有效改善算法性能。     

Complexity-aware algorithm architecture for real-time enhancement of local anomalies in hyperspectral images

Anomaly detection (AD) from remotely sensed multi-hyperspectral images is a powerful tool in many applications, such as strategic surveillance and search and rescue operations. In a typical operational scenario, an airborne hyperspectral sensor searches a wide area to identify regions that may contain potential targets. These regions typically cue higher spatial-resolution sensors to provide target recognition and identification. While this procedure is mostly automated, an on-board operator is generally assigned to examine in real time the AD output and select the regions of interest to be sent for cueing. Real-time enhancement of local anomalies in images of the over flown scene can be presented to the operator to facilitate the decision-making process. Within this framework, one of the ultimate research interests is undoubtedly the design of complexity-aware AD algorithm architectures capable of assuring real-time or nearly real-time in-flight processing and prompt decision making. Among the different AD algorithms developed, this work focuses on those AD algorithms aimed at detecting small rare objects that are anomalous with respect to their local background. One of such algorithms, called RX algorithm, is based on a local Gaussian assumption for background and locally estimates its parameters from each pixel local neighborhood. RX has been recognized to be the benchmark AD algorithm for detecting local anomalies in multi-hyperspectral images. RX decision rule has been employed to develop computationally efficient algorithms tested in real-time operating systems. These algorithms rely upon a recursive block-based parameter estimation procedure that makes their processing and, in turn, their detection performance differ from those of original RX. In this paper, a complexity-aware algorithm architecture fully adaptable to real-time processing is presented that allows the computational load to be reduced with respect to original RX, while strictly following its original formulation and thus assuring the same detection performance. An experimental study is presented that analyzes in detail the complexity reduction, in terms of number of elementary operations, offered by the proposed architecture with respect to original RX. A real hyperspectral image of a scene with deployed targets has been employed to perform a case-study analysis of the complexity reduction to be experienced in different operational scenarios. The real data are also adopted to illustrate a possible strategy for on-board line-by-line enhanced visualization of anomalies for decision support.     

Comparison of spectral and spatial windows for local anomaly detection in hyperspectral imagery

A new way of implementing two local anomaly detectors in a hyperspectral image is presented in this study. Generally, most local anomaly detector implementations are carried out on the spatial windows of images, because the local area of the image scene is more suitable for a single statistical model than for global data. These detectors are applied by using linear projections. However, these detectors are quite improper if the hyperspectral dataset is adopted as the nonlinear manifolds in spectral space. As multivariate data, the hyperspectral image datasets can be considered to be low-dimensional manifolds embedded in the high-dimensional spectral space. In real environments, the nonlinear spectral mixture occurs more frequently, and these manifolds could be nonlinear. In this case, traditional local anomaly detectors are based on linear projections and cannot distinguish weak anomalies from background data. In this article, local linear manifold learning concepts have been adopted, and anomaly detection algorithms have used spectral space windows with respect to the linear projection. Output performance is determined by comparison between the proposed detectors and the classic spatial local detectors accompanied by the hyperspectral remote-sensing images. The result demonstrates that the effectiveness of the proposed algorithms is promising to improve detection of weak anomalies and to decrease false alarms.     

基于GPU的图像特征并行计算方法

特征提取与描述是众多计算机视觉应用的基础。局部特征提取与描述因像素级处理产生的高维计算而导致其计算复杂、实时性差,影响了算法在实际系统中的应用。研究了局部特征提取与描述中的关键共性计算模块——图像金字塔机制及图像梯度计算。基于NVIDIA GPU/CUDA架构设计并实现了共性模块的并行计算,并通过优化全局存储、纹理存储及共享存储的访问方式进一步实现了其高效计算。实验结果表明,基于GPU的图像金字塔和图像梯度计算比CPU获得了30倍左右的加速,将实现的图像金字塔和图像梯度计算应用于HOG特征提取与描述算法,相比CPU获得了40倍左右的加速。该研究对于基于GPU实现局部特征的高速提取与描述具有现实意义。     

Unmixing-based content retrieval system for remotely sensed hyperspectral imagery on GPUs

This paper presents a new unmixing-based retrieval system for remotely sensed hyperspectral imagery. The need for this kind of system is justified by the exponential growth in the volume and number of remotely sensed data sets from the surface of the Earth. This is particularly the case for hyperspectral images, which comprise hundreds of spectral bands at different (almost contiguous) wavelength channels. To deal with the high computational cost of extracting the spectral information needed to catalog new hyperspectral images in our system, we resort to efficient implementations of spectral unmixing algorithms on commodity graphics processing units (GPUs). Spectral unmixing is a very popular approach for interpreting hyperspectral data with sub-pixel precision. This paper particularly focuses on the design of the proposed framework as a web service, as well as on the efficient implementation of the system on GPUs. In addition, we present a comparison of spectral unmixing algorithms available in the system on both CPU and GPU architectures.