A hybrid algorithm for subpixel detection in hyperspectral imagery

Broadwater, Joshua; Meth, Reuven; Chellappa, Rama

doi:10.1109/igarss.2004.1370633

Cited by 22 publications

(8 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since in MIL the positive bags are mixture of both true positive points and false positive points, the hybrid detector [45], [46] was introduced to the proposed objective function (4) to determine if instances from positive bags are the true positive points. Specifically, define the following detection statistic as an approximation of Pr(l ij = +|B i ),…”

Section: Multiple Instance Hybrid Estimatormentioning

confidence: 99%

Multiple instance hybrid estimator for hyperspectral target characterization and sub-pixel target detection

Jiao

Chen

McGarvey

et al. 2018

ISPRS Journal of Photogrammetry and Remote Sensing

View full text Add to dashboard Cite

The Multiple Instance Hybrid Estimator for discriminative target characterization from imprecisely labeled hyperspectral data is presented. In many hyperspectral target detection problems, acquiring accurately labeled training data is difficult. Furthermore, each pixel containing target is likely to be a mixture of both target and non-target signatures (i.e., sub-pixel targets), making extracting a pure prototype signature for the target class from the data extremely difficult. The proposed approach addresses these problems by introducing a data mixing model and optimizing the response of the hybrid sub-pixel detector within a multiple instance learning framework. The proposed approach iterates between estimating a set of discriminative target and non-target signatures and solving a sparse unmixing problem. After learning target signatures, a signature based detector can then be applied on test data. Both simulated and real hyperspectral target detection experiments show the proposed algorithm is effective at learning discriminative target signatures and achieves superior performance over state-of-the-art comparison algorithms.

show abstract

Section: Multiple Instance Hybrid Estimatormentioning

confidence: 99%

Multiple instance hybrid estimator for hyperspectral target characterization and sub-pixel target detection

Jiao

Chen

McGarvey

et al. 2018

ISPRS Journal of Photogrammetry and Remote Sensing

View full text Add to dashboard Cite

show abstract

“…This kind of method directly sets a threshold for the target abundance values of the pixels. Examples of this type of approach include the fully constrained least squares algorithm and representation methods [19,20,21]. The Sparse Representation-based Detector (SRD), originally developed for face recognition, has attracted considerable attention in the past ten years [22].…”

Section: Development Of Hsi Detection Algorithmsmentioning

confidence: 99%

Improving hyperspectral subpixel target detection using hybrid detection space

Latifi

2018

J. Appl. Rem. Sens.

View full text Add to dashboard Cite

A Hyper-Spectral Image (HSI) has high spectral and low spatial resolution. As a result, most targets exist as subpixels, which pose challenges in target detection. Moreover, limitation of target and background samples always hinders the target detection performance. In this thesis, a hybrid method for subpixel target detection of an HSI using minimal prior knowledge is developed. The Matched Filter (MF) and Adaptive Cosine Estimator (ACE) are two popular algorithms in HSI target detection. They have different advantages in differentiating target from background. In the proposed method, the scores of MF and ACE algorithms are used to construct a hybrid detection space. First, some high abundance target spectra are randomly picked from the scene to perform initial detection to determine the target and background subsets. Then, the reference target spectrum and background covariance matrix are improved iteratively, using the hybrid detection space. As the iterations continue, the reference target spectrum gets closer and closer to the central line that connects the centers of target and background and resulting in noticeable improvement in target detection. Two synthetic datasets and two real datasets are used in the experiments. The results are evaluated based on the mean detection rate, Receiver Operating Characteristic (ROC) curve and observation of the detection results. Compared to traditional MF and ACE algorithms with Reed-Xiaoli Detector (RXD) background covariance matrix estimation, the new method shows much better performance on all four datasets. This method can be applied in environmental monitoring, mineral detection, as well as oceanography and forestry reconnaissance to search for extremely small target distribution in a large scene. iv ACKNOWLEDGEMENT The first course I took at the University of Nevada, Las Vegas was "Coding with Applications in Computers and Communication Media" instructed by Dr. Shahram Latifi. This course triggered my enthusiasm in mathematics which had been concealed for many years. My master study at UNLV is memorable for me, for it is a valuable experience to learn some advanced knowledge in both engineering and math. I am grateful to many people who help me accomplish it. I would like to express my great gratitude to my advisor, Dr. Shahram Latifi. Without his help, my graduate study in the U.S. would not become true. He is a gracious professor and dedicated researcher. I would like to thank him for his encouragement, guidance and support during my master study. I would like to thank Dr. Yahia Baghzouz for his kind support and insightful instruction. I

show abstract

“…Finally, w is an additive Gaussian noise with zero mean and covariance Γ w ¼ σ 2 w I, representing both modeling and measurement errors. Based on the previous model, the AMSD is given by 21,22 …”

Section: Adaptive Matched Subspace Detectormentioning

confidence: 99%

Evaluation of the graphics processing unit architecture for the implementation of target detection algorithms for hyperspectral imagery

Trigueros-Espinosa

Vélez-Reyes

Santiago

et al. 2012

J. Appl. Remote Sens

View full text Add to dashboard Cite

Abstract. Recent advances in hyperspectral imaging sensors allow the acquisition of images of a scene at hundreds of contiguous narrow spectral bands. Target detection algorithms try to exploit this high-resolution spectral information to detect target materials present in a scene, but this process may be computationally intensive due to the large data volumes generated by the hyperspectral sensors, typically hundreds of megabytes. Previous works have shown that hyperspectral data processing can significantly benefit from the parallel computing resources of graphics processing units (GPUs), due to their highly parallel structure and the high computational capabilities that can be achieved at relative low costs. We studied the parallel implementation of three target detection algorithms (RX algorithm, matched filter, and adaptive matched subspace detector) for hyperspectral images in order to identify the aspects in the structure of these algorithms that can exploit the CUDA™ architecture of NVIDIA ® GPUs. A data set was generated using a SOC-700 hyperspectral imager to evaluate the performance and detection accuracy of the parallel implementations on a NVIDIA ® Tesla™ C1060 graphics card, achieving real-time performance in the GPU implementations based on global statistics.

show abstract

A hybrid algorithm for subpixel detection in hyperspectral imagery

Cited by 22 publications

References 7 publications

Multiple instance hybrid estimator for hyperspectral target characterization and sub-pixel target detection

Multiple instance hybrid estimator for hyperspectral target characterization and sub-pixel target detection

Improving hyperspectral subpixel target detection using hybrid detection space

Evaluation of the graphics processing unit architecture for the implementation of target detection algorithms for hyperspectral imagery

Contact Info

Product

Resources

About