Research On Target Detection Of Hyperspectral Image Based On Visual Saliency

With the advancement of technology and the popularization of a large number of electronic devices,people have higher and higher requirements for the quality of images.Then,hyperspectral image has emerged as the times require.Hyperspectral image contains a large amount of spectral data,which is very helpful for target detection.However,too much spectral data causes difficulties for image processing.At present,the target detection of hyperspectral image has become an important research topic.It is of great significance for some research such as target classification,image segmentation,and target tracking.The salient target can be quickly selected by using visual saliency detection,which greatly improves the efficiency and accuracy of target detection.Therefore,it has certain research value and significance to introduce the visual saliency detection model into the target detection task of hyperspectral images.Based on this,this paper proposes a method for detecting salient target in hyperspectral image based on visual saliency.The main research contents are as follows:1.Aiming at the problem of incomplete retention of target frequency in the salient target detection of hyperspectral image,resulting in poor target result and unclear edges.This paper proposes a hyperspectral image target detection algorithm based on frequency tuning saliency detection.First,the frequency-tuning saliency detection algorithm is used to detect the salient target of the hyperspectral image,and generate a saliency map with clearly defined boundaries.At the same time,make full use of the spectral data contained in the hyperspectral image as an information supplement and to calculate the distance between the spectral vectors.Then,the spectral saliency map is obtained which to improve the accuracy of target detection.Finally,the two saliency maps are merged to form the final saliency target map.Experimental results show that our method is superior to other hyperspectral image saliency target detection algorithms in visual effects,and it also performs well in the evaluation of the precise-recall curve and F-measure measurement.2.Aiming at the problem that in the case of complex background,some unfavorable factors cause interference to the salient target detection of hyperspectral image,a method of hyperspectral image target detection based on global contrast saliency is proposed.First,the global contrast saliency detection algorithm is used to detect the hyperspectral image.By spatially weighting the region,the salient target map with suppressed background can be quickly obtained.At the same time,the spectral saliency map is obtained by calculating the spatial distribution of the spectral information,so as to improve the accuracy of target detection.Finally,the two saliency maps are merged to obtain a result map.Through qualitative comparison and quantitative evaluation of this method with other methods,the effectiveness and superiority of this method are proved.