基于轮廓合成的人耳图像边缘检测

Ear Image Edge Detection Based On Contour Composition

人耳识别技术作为一种新的研究在生物特征识别领域提出一种新思路。边缘检测是图像特征分析识别的基础,检测质量直接影响系统识别率。本文针对人耳图像的局部特征和识别要求提出一种基于轮廓合成的边缘检测方法。该算法对采集到的图像预处理,将Sobel算子与阈值法得出的边缘图像合成,最后根据统计结果用去除干扰技术处理非边缘。这种方法得出的耳朵边缘图像清晰完整无干扰,比单一采用其他方法效果好,为后续特征提取及识别提供较好基础。

Research of ear recognition technology creates a new way in the field of biometrics recognition. Edge detection is the base of extraction and analysis. The quality of detection will directly affect the final recognition rate. According to the feature of ear image and requirement of extraction, this paper proposes a new edge detection method based on contour composition. This arithmetic pretreats the ear image and then composes the useful edge of Sobel arithmetic operator method and threshold method. The non-edge line will be erased by use of disturbed line removed technique according to the result of statistic. Clear, connected and non-disturbed ear edge image got from this method will be offered to image feature extraction. Obviously, this method is better than others which only use edge detection arithmetic operators and provides better edge image for extraction and analysis of ear features.