Bulent Ayhan

Anomaly detection in hyperspectral images through spectral unmixing and low rank decomposition

Anomaly detection has been known to be a challenging, ill-posed problem due to the uncertainty of anomaly and the interference of noise. In this paper, we propose a novel low rank anomaly detection algorithm in hyperspectral images (HSI), where three components are involved. First, due to the highly mixed nature of pixels in HSI, instead of using the raw pixel directly for anomaly detection, the proposed algorithm applies spectral unmixing algorithms to obtain the abundance vectors and uses these vectors for anomaly detection. Second, for better classification, a dictionary is built based on the mean-shift clustering of the abundance vectors to better represent the highly-correlated background and the sparse anomaly. Finally, a low-rank matrix decomposition is proposed to encourage the sparse coefficients of the dictionary to be low-rank, and the residual matrix to be sparse. Anomalies can then be extracted by summing up the columns of the residual matrix. The proposed algorithm is evaluated on both synthetic and real datasets. Experimental results show that the proposed approach constantly achieves high detection rate while maintaining low false alarm rate regardless of the type of images tested.

Pansharpening of Mastcam images

This paper summarizes a new investigation of applying advanced pansharpening algorithms to enhance the images of the left imager in the Mastcam onboard the Curiosity rover, which landed on Mars in 2012. The various instruments on the rover have already made great contributions in the understanding of Mars. The goal of our research is to generate both high spatial and high spectral image cube by using the left and right Mastcam imagers. Eleven algorithms have been investigated using five objective performance metrics. Subjective evaluations have also been conducted. The image enhancement results are encouraging.

Enhancing Mastcam Images for Mars Rover Mission

This paper summarizes some new results in improving the left Mastcam images of the Mars Science Laboratory (MSL) onboard the Mars rover Curiosity. There are two multispectral Mastcam imagers, having 9 bands in each. The left imager has wide field of view, but low resolution whereas the right imager is just the opposite. Our goal is to investigate the possibility of fusing the left and right images to form high spatial resolution and high spectral resolution data cube so that stereo images and data clustering performance can be improved. Many pansharpening algorithms have been investigated. Actual Mastcam images were used in our experiments. Preliminary results indicate that the pansharpened images can indeed enhance the data clustering performance using both objective and subjective evaluations.

A Novel Utilization of Image Registration Techniques to Process Mastcam Images in Mars Rover With Applications to Image Fusion, Pixel Clustering, and Anomaly Detection

The Mars Science Laboratory is a robotic rover mission to Mars launched by NASA on November 26, 2011, which successfully landed the Curiosity rover in Gale Crater on August 6, 2012. The Curiosity rover has two mast cameras (Mastcams) that acquire stereo images at a number of different wavelengths. Each camera has nine bands of which six bands are overlapped in the two cameras. These acquired stereo band images at different wavelengths can be fused into a 12-band multispectral image cube, which could be helpful to guide the rover to interesting locations. Since the two Mastcams’ fields of view are three times different from each other, in order to fuse the left- and right-camera band images to form a multispectral image cube, there is a need for a precise image alignment of the stereo images with registration errors at the subpixel level. A two-step image alignment approach with a novel utilization of existing image registration algorithms is introduced in this paper and is applied to a set of Mastcam stereo images. The effect of the two-step alignment approach using more than 100 pairs of Mastcam images, selected from over 500000 images in NASAs Planetary Data System database, clearly demonstrated that the fused images can improve pixel clustering and anomaly detection performance. In particular, registration errors in the subpixel level are observed with the applied alignment approach. Moreover, the pixel clustering and anomaly detection performance have been observed to be better when using fused images.

Speech separation algorithms for multiple speaker environments

Conventional speaker identification and speech recognition algorithms do not perform well if there are multiple speakers in the background. For high performance speaker identification and speech recognition applications in multiple speaker environments, a speech separation stage is essential. Here we summarize the implementation of three speech separation techniques. Advantages and disadvantages of each method are highlighted, as no single method can work under all situations. Stand-alone software prototypes for these methods have been developed and evaluated.

An integrated approach to robust speaker identification and speech recognition

Conventional speaker identification and speech recognition algorithms cannot deal with noisy and multiple speaker environments. For example, IBM via Voice has low recognition rates if dictation is done in a noisy environment. In order to achieve high performance in speaker identification and speech recognition, we propose an integrated approach that takes every facet of the process into account. Here we summarize some preliminary results from the application of this integrated approach to robust speaker identification and speech recognition. A real-time stand-alone software prototype has been developed to evaluate the effectiveness of the approach.

Application of support vector machines to vapor detection and classification for environmental monitoring of spacecraft

Electronic noses (E-nose) have gained popularity in various applications such as food inspection, cosmetics quality control [1], toxic vapor detection to counter terrorism, detection of Improvised Explosive Devices (IED), narcotics detection, etc. In the paper, we summarized our results on the application of Support Vector Machines (SVM) to gas detection and classification using E-nose. First, based on experimental data from Jet Propulsion Lab. (JPL), we created three different data sets based on different pre-processing techniques. Second, we used SVM to detect gas sample data from non-gas background data, and used three sensor selection methods to improve the detection rate. We were able to achieve 85% correct detection of gases. Third, SVM gas classifier was developed to classify 15 different single gases and mixtures. Different sensor selection methods were applied and FSS & BSS feature selection method yielded the best performance.

A Comparative Study of Spatial Speech Separation Techniques to Improve Speech Recognition

Robust speech recognition in noisy and reverberant conditions is an important research area in recent years. Here we present a comparative study of several spatial speech separation methods. The main performance metric is word error rate (WER) under different signal-to-noise ratio (SNR) and reverberant conditions. Extensive simulations showed that one technique known as polyaural processing stood out as the best one.

Robust Speaker Identification Algorithms and Results in Noisy Environments.

In this research, we have developed a robust speaker identification system, which involves mask estimation, gammatone features with bounded marginalization to deal with unreliable features, and Gaussian mixture model (GMM) for speaker identification. Extensive experiments using actual and synthesized conversations clearly demonstrated the performance of our algorithms under noisy conditions.

On the generation of high-spatial and high-spectral resolution images using THEMIS and TES for Mars exploration

In the 2015 NASA ROSES solicitation, NASA has expressed strong interest in improving the accuracy of Mars surface characterization using satellite images. Thermal Emission Imaging System (THEMIS), an imager with a spatial resolution of 100 meters, has 10 infrared bands between 6 and 15 micrometers. Thermal Emission Spectrometer (TES), an imager with a spatial resolution of 3 km, has 143 bands between 5 and 50 micrometers. While both imagers have a variety of applications, it would be ideal to generate high-spatial and high-spectral resolution data products by fusing their respective outputs. We present a novel approach to fusing THEMIS and TES satellite images, aiming to improve orbital characterization of Mars’ surface. First, the THEMIS bands must undergo atmospheric compensation (AC) due to the presence of dust, ice, carbon dioxide, etc. A systematic AC procedure using elevation information and spectrally uniform pixels has been developed and implemented. Second, a set of proven pan-sharpening algorithms has been applied to fuse the two sets of images. The pan-sharpened images have the spatial resolution of THEMIS images and the spectral resolution of TES images. The results of extensive experiments using THEMIS and TES data collected near the Syrtis Major region (one of the final 3 candidate landing sites for the Mars 2020 rover) clearly demonstrate the feasibility of the proposed approach.