一种新的二值图像目标轮廓跟踪算法  

A New Algorithm on Objects Boundary Tracking of Binary Image

作　　者：周丰乐[1] 徐向民[1] 肖跃[1] 周娇[1] 

机构地区：[1]华南理工大学电子与信息学院,广东广州510640

出　　处：《微计算机信息》2007年第02X期259-261,共3页Control & Automation

基　　金：广东省自然科学基金资助(31449;05100514);广东省科技攻关项目(2005B10201052)

摘　　要：提出了根据方向记忆和动态权值确定下一边界点的二值图像目标轮廓边界跟踪算法。首先搜索轮廓内边界予以标记,并以该标记作为指导,结合上一边界点位置对搜索方向动态赋予不同权值,综合分析后得到下一轮廓边界点。由于采用内外边界的对照搜索,可以较好地过滤图像二值化引起的边界噪声。对于不封闭曲线,本算法也可通过一次扫描得到其轮廓信息。实验表明,该算法在轮廓边界提取速度和准确度方面都有较以往跟踪算法更好的性能,在搜索智能化和处理复杂轮廓方面也有更好的表现。A novel boundary tracking algorithm of binary image according to the direction memory and dynamic weight is proposed, First, the inner boundary of the contour is searched and marked.Guided by the marker of the inner boundary, considered the previous boundary point＂ s position, the dynamical Jirection weights are calculated.Then the next boundary point is selected under synthesized analysis.The boundary noise caused by single threshold of the binary image is well filtrated because of the parallel search of both inner and outer boundary.The experiment result indicates that this new algorithm has better performance than other tracking algorithms both in the aspect of the contour extraction speed and precision.

关 键 词：二值图像 边界跟踪 方向记忆 动态权值 边界去噪 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]