Abstract: For the target camouflage evaluation in hyperspectral reconnaissance, a novel evaluation method of hyperspectral camouflage effect is proposed based on detection and perception theory. The local anomaly detection operator is adopted to extract the difference value among each pixel and the surrounding pixels. On this basis, the potential target areas are acquired using the space density clustering and neighborhood fusion algorithm. The significant indicators reflecting the spectral differences and overall camouflage characteristics are established, and a new idea about hyperspectral camouflage evaluation based on the limited-time search strategy is proposed to evaluate the camouflaged targets from multiple aspects, such as search time and evaluation score. Simulated results show that the proposed method is used to overcome the problem of single evaluation index of traditional evaluation method based on multi-feature description, and give the objective and accurate evaluation conclusion. Key

Key words: hyperspectralimage, camouflageevaluation, anomalydetection, limited-timesearch, visualattentionmodel