结合边缘检测和区域分割的形状特征提取

提出了一种将Canny边缘检测和区域分割相结合的形状特征提取方法.该方法通过区域分割得到的粗略形状轮廓,过滤掉边缘检测轮廓中不规则或冗余的边界点,得到位置比较准确的主轮廓图像,减少形状描述的计算量.为验证方法的有效性,建立了一个基于形状特征的图像检索系统框架,实验结果表明,该方法提高了系统检索准确性.

Shape Feature Extraction Based on Edge Detection and Region Segmentation

An algorithm for shape feature extraction based on Canny edge detection and region-based segmentation is proposed in the paper. In this algorithm,many of the edge points obtained by Canny edge detection are anomalous and redundant. To solve this problem,a rough sketch is obtained by region segmentation and used to filter the Canny edge detection contour line,thereby the primary figure of an image may be characterized by the method. The primary figure has accurate position,and can decrease the amount of calculation when describing the shape feature. In order to verify the effectiveness of the method,a CBIR prototype is built. Experimental results show that the method improves the search accuracy of the system.