基于流场纹理表达与表面粗糙度测量的运动目标分割  

Moving target segmentation based on flow field texture representation and surface roughness measurement

作　　者：张旭光[1] 贺海明[1] 苏连成[1] 李小俚[1] 

机构地区：[1]燕山大学电气工程学院.河北秦皇岛066004

出　　处：《燕山大学学报》2013年第4期353-357,共5页Journal of Yanshan University

基　　金：国家自然科学基金资助项目(61271409,60905046);河北省自然科学基金资助项目(F2013203364);中国博士后科学基金资助课题(2012M510768,2013T60264);中国留学基金资助课题(2011813018)

摘　　要：针对传统的光流法和背景差法所分割的运动目标存在区域丢失和内部孔洞的缺点,本文提出一种基于流场纹理表达及物体表面粗糙度测量的运动目标分割方法,即通过计算相邻帧间的光流将目标运动表达成流体的流动,并利用线积分卷积来表达流场纹理以展示更多的流场细节,从而将背景图像和运动目标分别表达成不同的纹理,最后通过针描法测量物体表面粗糙度的策略将运动目标和背景的纹理图像映射为表面粗糙度不同的物体,并通过分析直方图确定阈值分割运动目标。实验结果表明,本文提出的运动目标分割方法可自适应地选取阈值,并且可克服内部孔洞。Aiming at the drawback of region impairment and inside holes detected by background subtraction and optical flow based method, a moving target segmentation method based on flow field representation and surface roughness measurement is proposed. The motions of targets are described by the flow of fluid by computing the optical flow between two consecutive frames. A Line integral convolution method is used to represent the textures to reveal more details of flow field. Therefore, the moving targets and background regions are expressed by different texture images. Finally, the textures of moving target and background regions are mapped as objects with different surface roughness by the strategy of surface roughness measurement with stylus probe method. Furthermore, the threshold for segmentation the surface roughness image is gained by analyzing the histogram. Experiment results show that the proposed method can select the threshold adaptively, and can overcome the problem of inside holes.

关 键 词：运动目标分割 光流场 线积分卷积 表面粗糙度测量 针描法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]