Polarity SAR target detection method based on FCN-CRF master-slave network

Abstract

The invention provides a polarity SAR target detection method based on a FCN-CRF master-slave network. The method comprises steps of inputting a to-be-detected polarity SAR image, and carrying out delicate polarity Lee filtering on a polarity coherence matrix T of the polarity SAR image to filter coherent noise so as to obtain the filtered coherence matrix, wherein each element of the filtered coherence matrix is a 3*3 matrix, that is to say, each pixel point has nine-dimensional features. According to the invention, by expanding image block features into pixel-level features, the correlation degree of selected training samples through matching of pixel points of a region of interest is quite high and quite effective; the feature image blocks with the pixel points of the region of interest whose quantity is less than 50% of the whole image block will not participate in following calculation, so the operand is greatly reduced and the detection efficiency is improved; by using the Lee filtering to pre-process of the original polarity SAR image, coherence spot noise is effectively reduced and image quality and detection performance are improved; and by use of spiral scattering components corresponding to urban buildings obtained through the Yamaguchi decomposition, features of polarity SAR artificial targets are effectively extracted, and detection precision of the artificial targets is improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a polarization SAR target detection method based on an FCN-CRF master-slave network. Background technique [0002] Polarization SAR has the advantages of all-weather, all-time, high resolution, side-view imaging, etc. It can obtain the fine features and geometric features of the target. It is very urgent to make fast and accurate detection of man-made targets in military and civilian applications. The efficient use of convolutional network in image feature extraction makes it have important theoretical value and broad application prospects in solving the problem of extremely SAR artificial target detection. [0003] Traditional convolutional networks such as CNN use image blocks as input for training and prediction to implement a pixel category detection task, which not only increases storage space, but also has low computational efficiency, and the size of ...

AI Technical Summary

Problems solved by technology

[0003] Traditional convolutional networks such as CNN use image blocks as input for training and prediction to implement a pixel category detection task, which not only increases storage space, but also has low computational efficiency, and the size of image blocks is much smaller than the entire image. Lead to the loss of some features, thus limiting the detection performance

[0004] In 2015, in response to the problems existing in CNN, Jonathan Long et al. proposed Fully Convolutional Networks, referred to as FCN. This network extends the category detection task from the image level to the pixel level, thereby detecting the region of interest, but the FCN detection result is not fine enough. It is easy to ignore the image details, and the pixel-level detection fails to fully consider the spatial neighborhood information

Images