Research Of Anomaly Detection Algorithms Of Hyperspectral Imagery Based On Kernel Method

Hyperspectral imagery which can distinguish different ground objects and have tiny spectral difference by virtue of high spectral resolution is a new type of remote sensing data.Hyperspectral imagery Also it doesn't need any prior knowledge of the target spectral signature. Thus it is very practicable in real scenes. Nowadays it becomes a hotspot in the field of target detection, and attracts many specialists and scholars'attention. Based on analysing of structure and characteristics of hyperspectral imagery, and applying some signal processing techniques, the dissertation does the following researches in order to solve the difficulties in anomaly detection, such as the single detection mode, the choosen of the appropriate kernel parameter, and using the spatial correlation.Firstly, after researching on the detection mode of the anomaly detection for hyperspectral imagery, a new KRX algorithm based on multiple-window algorithm analysis is introduced to slove the problems of some interference signal source which are unknown, or not interesting. The algorithm uses three windows to detect for hyperspectral imagery which are the outer window, the middle window and the inner window. The outer window removes the white noise which comes from the middle window and the inner window and diminishes the influence of background interference by the use of OSP algorithm. The background pixel selected by middle window detects the pixel selected by inner window by using the KRX algorithm. The results prove that the proposed algorithm outperforms the other algorithms, and can obtain a better effect of detection and a lower false alarm rate.Secondly, the theory of KOSP is analyzed and a new target orthogonal subspace projection anomaly detection algorithm for hyperspectral image based on adaptive kernel method (AKOSP) is introduced. This algorithm solves that detection parameter is difficult to adapt to complex and changing background environment which declines the detection efficiency. This algorithm not only enhanced the universality also reduces the work of testing. The results prove that the proposed algorithm outperforms the other algorithms, and can obtain a better detection result to prove that the kernel parameter which is determined appropriately or not is an important factor in the decision algorithm performanceFinally, In the problems of previous hyperspectral target detection algorithms are based on the information of spectrum and the data of feature space, but the spatial correlation is neglected. A new hyperspectral target detection algorithm based on digital image morphological theory is devised to cope with this. Firstly, reduce the dimensions by the use of the morphological closing transform theory. Then detect the hyperspectral Imagery reduced the dimensions by the ACO-KRX operator. This operator is not only supported by the information of spectrum, but by the spatial correlation. So ACO-KRX algorithm can diminish the influence of noise interference and smooth the image.