Bayesian saliency based SAR image target detection method

Abstract

The invention discloses a Bayesian saliency based SAR image target detection method and mainly solves the problems of low detection accuracy and incomplete structure of a detected target on a binary image in an existing SAR image target detection technology. The method is implemented by the steps of performing superpixel segmentation on an original SAR image; constructing a prior saliency map, a foreground likelihood graph and a background likelihood graph by utilizing a superpixel segmentation result; fusing results of the prior saliency map, the foreground likelihood graph and the background likelihood graph in a Bayesian framework to obtain a Bayesian posterior saliency map; segmenting the Bayesian posterior saliency map to obtain a binary image with a suspected target region; and performing clustering and false alarm region removal on the binary image and extracting a suspected target slice on an original SAR image, thereby finishing the SAR image target detection. Compared with dual-parameter CFAR detection, the method has the advantages of high detection accuracy and relatively complete structure of the detected target on the binary image, thereby being suitable for SAR image target detection in complicated scenes.

Description

technical field [0001] The invention belongs to the field of radar target detection, in particular to a SAR image target detection method, which can be used for ground vehicle target detection. Background technique [0002] Radar imaging technology was developed in the 1950s, and has been developed by leaps and bounds in the next 60 years. At present, it has been widely used in military, agriculture, forestry, geology, ocean, disaster, mapping and many other fields. [0003] Synthetic Aperture Radar (SAR) is an active sensor that uses microwaves for perception. Compared with other sensors such as infrared and optical, SAR imaging is not limited by conditions such as light and weather. observation. Therefore, SAR has become an important means of earth observation and military reconnaissance, and the automatic target recognition technology of SAR images has attracted more and more attention. [0004] SAR automatic target recognition ATR technology usually adopts the three-le...

AI Technical Summary

Problems solved by technology

Although two-parameter CFAR is a widely used classic SAR image target detection algorithm, there are two main problems: first, because the statistical distribution model of background clutter does not necessarily obey the Gaussian distribution, especially in SAR images in complex scenes, The inaccurate estimation of the parameters of the clutter statistical model in this target detection method makes it prone to false alarms and missing alarms, and the target detection accuracy is low; secondly, since the two-parameter CFAR is a pixel-level target detection method, it does not consider The structure information of the target, the target pixels on the detected binary image are relatively discrete and the structure is incomplete, resulting in a large number of slices that only cut part of the target in the suspected target slices extracted by this target detection method, and most of the targets are not located in the In the center of the slice, the object detection accuracy is low

Images