基于复合感受野的轮廓检测算法ContourDetectionBasedon-Core.PDF

第31 卷第7 期 电 子 与 信 息 学 报 Vol.31No.7 2009 年7 月 Journal of Electronics Information Technology Jul..2009 基于复合感受野的轮廓检测算法 杜晓凤① 李翠华② 李 晶③ ①(厦门大学自动化系 厦门 361005) ②(厦门大学计算机系 厦门 361005) ③(北京邮电大学软件学院 北京 100876) 摘 要：自然场景自动检测目标轮廓是非常困难的。该文提出了一种基于复合感受野的图像轮廓检测仿生模型。在 传统的基于抑制模型的基础上引入去抑制机制，在减少背景纹理干扰的同时，保护较弱的轮廓，从而尽量保持轮廓 的完整性。实验结果表明，这种仿生模型可以有效地抑制纹理边缘，减少对轮廓的破坏，提高了自然背景中轮廓检 测的性能。 关键词：目标识别；轮廓检测；初级视皮层；复合感受野 中图分类号：TP391.41 文献标识码：A 文章编号：1009-5896(2009)07-1630-05 Contour Detection Based on Compound Receptive Field ① ② ③ Du Xiao-feng Li Cui-hua Li Jing ① (Department of Automation, Xiamen University, Xiamen 361005, China) ② (Department of Computer Science, Xiamen University, Xiamen 361005, China) ③ (School of Software Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China) Abstract: Contour detection from natural scenes is a tough task in computer vision. This paper proposes a visual model of contour detection based on compound receptive field. By inducing dis-inhibition mechanism to traditional model that was based on inhibition mechanism, inferences are suppressed while weak edges are preserved, so that the integrity of contour is improved. Experiment results show that this approach can suppress edges due to texture, protect object contour and improve the performance of contour detection from nature scene. Key words: Target identification; Contour detection; Primary visual cortex; Compound Receptive Field(CRF)