中文摘要：

轮廓是物体的一个重要特征,轮廓提取的好坏对后续的目标定位、特征提取、识别和分类有着重要的影响.文中在现有方法的基础上,提出了一种新的轮廓提取方法.首先,在经典Otsu算法基础上,结合图像的熵给出了一种图像分割阈值选取方法;然后通过划分单元块,统计非零像素个数,去除噪声块,合并单元块,再次去除噪声块获得目标区域;最后采用4-邻域方法提取目标区域的轮廓.实验结果显示,改进的算法能够有效分割出目标,设计的目标轮廓提取方法能够有效去除噪声的影响,获得封闭的、单像素宽度的外轮廓线,为后续的图像处理和目标识别奠定了良好的基础.

英文摘要：

As contour is an important feature of the object,and its extraction quality has a significant influence on the following works like object location,feature extraction,recognition and classification. In this paper,a new method of contour extraction is proposed based on the existing methods. Firstly,the methods of threshold selection and segmentation is presented,which is based on Otsu algorithm and entropy of image. Then we get the target area by the following steps： diving unit blocks,counting the non-zero pixel number,removing the noise blocks,merging unit blocks,and removing the noise blocks again. Finally,the contour of the target region is extracted by using the method of the 4-neighborhood. Experimental results show that the improved algorithm can effectively segment the target and the designed contour extraction algorithm will remove the noise very well,which extract a closed,single pixel width of the contour,laying a good foundation for future image processing and target recognition.