基于引力场增强的SAR图像舰船检测方法研究

船只检测是实现船只航行安全的重要措施之一,利用SAR图像可实现船只检测。然而,传统的一些方法一般容易受到SAR图像斑噪的影响,在检测结果中产生大量的虚警。为解决这一问题,本文提出了一种基于引力场增强的舰船检测方法。该方法利用像素与其邻域内像素的相互作用可对目标像素增强的效应,有效地抑制了斑噪像素和背景像素的强度,凸显了目标。由于增强后的像素已经不满足对海面区域的均质性假设,因此直接使用恒虚警检测算法对图像进行全局检测并不能够得到很好的效果,据此本文引入了一个基于均质区域自适应分割的改进的K-CFAR检测算法,将图像分割为不同大小的一系列均质区域,并分别对各个均质区域使用一个改进的K-CFAR检测器对船只目标进行检测。最后,使用Radarsat-1数据和Envisat ASAR数据对本文算法进行了验证。实验表明,本文提出的方法能够有效地凸显弱目标,增加检测准确性,降低检测的虚警概率。

A Ship Detection Method in SAR Image Based on Gravity Enhancement

Automatic ship detection with SAR imagery is one of the important applications with SAR data,and it is widely used in many areas,such as fishery monitoring,environment protection,coastal security and military application.A large amount of ship targets detection researches using SAR data have been conducted since 1980s,and many detection techniques have been proposed.However,most of the traditional ship detection methods can be affected by the SAR speckle and background clutter,and many false alarms may be introduced in the detection results.To improve the ship detection capability,a new ship detection method,which is based on the gravity enhancement,is proposed in this paper.Initially,the targets pixels are enhanced by the interactions between themselves and their neighbors,and it is proved that SAR speckle and background clutter will be suppressed after the enhancement as well as the contrasts between targets and clutter have been greatly increased.Then,as the enhanced SAR imagery may not satisfy the homogenous hypothesis,an improved K-CFAR detector based on an adaptive segmentation algorithm is applied to the enhance imagery.The adaptive segmentation algorithm segments the whole marine area into a series of homogenous sub-images by the homogenous test,and the improved K-CFAR detector,which can adaptively choose probability density function(pdf),will be applied to each sub-image to detect the ship targets.Finally,some Radarsat-1 data and Envisat ASAR data are used to validate the algorithm mentioned in this paper.Experimental results demonstrate that the ship detection algorithm based on gravity enhancement can effectively enhance ship targets as well as suppress speckle and clutter,therefore,detection rate will be increased and the false alarm rate will be reduced.