GatorSense 高光谱图像分析工具包_MATLAB_代码_下载

# Gatorsense hsitoolkit signature detectors MATLAB version hsitoolkit/signature_detectors Demo instructions: (1) load an_hsi_img_for_tgt_det_demo.mat (2) run `det_out = signature_det_demo(hsi_sub, tgt_spectra, [], wavelengths, gtImg_sub)` Suite of signature detectors implemented so far: - abd_detector: Abundance of target signature when unmixed using target signature and background endmembers assuming the linear mixing model - ace_detector: Squared Adaptive Cosine/Coherence Estimator (Squared ACE), cosine of vector angle between target and pixel spectra after whitening based on background statistics, squared - ace_local_detector: Adaptive Cosine/Coherence Estimator where background statistics are estimated from local window - ace_ss_detector: Squared Adaptive Cosine/Coherence Estimator Subspace Formulation - acert_detector: Adaptive Cosine/Coherence Estimator (ACE), cosine of vector angle between target and pixel spectra after whitening based on background statistics - acert_max_detector: ACE given multiple target signatures. Confidence value for each pixel is max ACE score over all target signatures. - amsd_detector: Adaptive Matched Subspace Detector - ccmf_detector: Class Conditional Matched Filter, Segment data using a Gaussian Mixture Model and then apply SMF to each component - cem_detector: Constrained Energy Minimization Detector - ctmf_detector: Cluster Tuned Matched Filter - fam_statistic: False Alarm Mitigation Statistic - ftmf_statistic: Finite Target Matched Filter - ha_detector: Hybrid Abundance Detector, unmix using background endmembers as well as using background and target endmembers, model proportions with a Gaussian mixture, compute pixel-wise likelihood ratios - hs_detector: Hybrid Subpixel Detector - hsa_detector: Hybrid Structured/Abundance Detector - hsd_detector: Likelihood ratio after unmixing with background and unmixing with background and target signature (Broadwater and Chellappa's method) - hsd_local_detector: Hybrid Structured Detector using local background estimation - hua_detector: Hybrid Unstructured Abundance Detector - hud_detector: Hybrid Unstructured Detector (Broadwater and Chellappa's method) -mtmf_statistic: Mixture Tuned Matched Filter Infeasibility Statistic - osp_detector: Orthogonal Subspace Projection Detector - palm_detector: Pairwise Adaptive Linear Matched Filter - qmf_detector: Quadratic Spectral Matched Filter - sam_detector: Spectral Angle Mapper, calculates vector angle between target signature and each pixel spectrum - smf_detector: Spectral Matched Filter, inner product between target and pixel spectra after whitening based on background statistics - smf_local_detector: Spectral Matched Filter using local background statistics - smf_max_detector: Spectral Matched Filter given multiple target signatures. Confidence value for each pixel is max SMF score over all target signatures. - spsmf_detector: Subpixel Spectral Matched Filter Signature detectors can also be run in *Segmented* mode using segmented.m utility. Segmented mode is where a detector is applied to segments of the imagery separately (i.e., background statistics computed from segment rather than full image). See segmented examples in demo code. Review and references for many of these signature detectors can be found in the literature review in Taylor Glenn's Ph.D. thesis: Glenn, Taylor C. Context-dependent detection in hyperspectral imagery. Diss. University of Florida, 2013. Contact: Alina Zare, [email protected]