基于灰度分布模型的声呐图像目标检测算法  

Novel target detection algorithm of sonar image based on gray distribution model

作　　者：田晓东[1] 刘忠[1] 

机构地区：[1]海军工程大学电子工程学院,湖北武汉430033

出　　处：《系统工程与电子技术》2007年第5期695-698,共4页Systems Engineering and Electronics

摘　　要：针对声呐图像目标检测问题,提出了一种基于声呐灰度图像像素点的灰度分布模型的目标检测快速算法。该算法利用图像中各像素点处滑动窗内像素点的灰度分布模型的参数和拟合误差为特征量,构造声呐图像的特征图,并采用自适应阈值算法进行目标特征区域检测处理。最后将该算法的检测结果同分形特征以及扩展分形特征检测结果进行了比较分析。仿真结果表明该算法具有实时性好、准确度高的特点,并且可有效地克服背景中演示等自然景物的影响,实现人造目标的准确提取。Aiming at terms of target detection of the high-frequency sonar image, and a novel algorithm based on analysis of pixel＇s grey-level distribution function in the sonar image, used to target detection, is presented. In the regions of sliding window around each pixel in sonar image, distribution function parameters and matching error are estimated, which are adopted to construct characteristic images of sonar image. Adaptive threshold algorithm is used to detect target fast and exactly. In the end, simulation results of this method presented are compared with those from fractal feature and extended fractal feature. Simulation results indicate that this method has characteristics of real-time and high-precise. Besides, it can overcome the influence of natural objects such as rock to extract man-made target.

关 键 词：声呐图像 目标检测 灰度分布模型 拟合误差 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]