录播系统中图像分割与轮廓提取算法研究  

Research on Image Segmentation and Contour Extraction in Video Recording and Broadcasting System

作　　者：赵以国[1] 唐贵进[1] 朱秀昌[1] 

机构地区：[1]南京邮电大学江苏省图像处理与图像通信重点实验室,江苏南京210003

出　　处：《电视技术》2015年第13期32-35,共4页Video Engineering

基　　金：国家自然科学基金项目(61071091)

摘　　要：在基于图像识别技术的录播系统中,通常使用背景差分法检测运动目标,在当前帧与背景帧做差之后,需要进一步选取合适的阈值进行图像分割,从而最终提取出运动目标的轮廓。研究了基于最大类间方差准则的图像阈值分割,在此基础上进一步研究了掏空内部点的轮廓提取算法,并且使用轮廓缺陷修补技术改进轮廓提取算法,从而得到封闭的轮廓,最后使用Freeman链码表示提取的轮廓。实验证明,该算法为录播系统的实现提供了良好的技术保障。The recording and broadcasting system based on image recognition technology usually uses the background subtraction algorithm to detect the moving object. After making the difference between the current frame and the background frame, it needs to select the appropriate threshold for image segmentation in order to extract the moving object＇ s contour. The image threshold segmentation is studied based on Otsu criterion, and the contour extraction algorithm is further studied based on the hollowed interior points. The technology of contour defect repair is used to improve the contour extraction algorithm, then the closed contour is got. Finally, the Freeman chain code represent the contour is used. Experimental results show that the proposed algorithm provides a good technical support to realize the recording and broadcasting system.

关 键 词：录播系统 图像分割 最大类间方差 轮廓提取 FREEMAN链码 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]