Abstract: Color edge detection based on CNN has attracted many researches. Most existing works utilize fixed threshold based on experience to design CNN templates. However, this method ignores the fact that the human vision system has strong adaptability. Adaptive thresholds are brought into the progress of designing CNN templates, which consider high adaptability of human vision system. Meanwhile, the stability of this algorithm is comprehensively discussed. The algorithm is able to make detected edges be more acceptable by the human vision system. Performance evaluation results validate the efficiency of the proposed algorithm.

Key words: edge detection, cellular neural networks, threshold, adaptability

摘要： 细胞神经网络用于彩色图像边缘提取已经有很多人做了研究。现有的大部分工作都根据经验选取固定阈值来设计CNN模板。但这种阈值的选取方法忽略了人眼最小分辨差具有自适应性的特点。在设计图像边缘提取CNN模板选取阈值时，引入人眼最小阈值差成果，设计出了一组阈值自适应的CNN模板，同时对设计的阈值自适应算法的稳定性进行详细的论证。该算法让检测出的边缘更加符合人眼的视觉特性。实验结果证明，该算法效果良好。

关键词: 边缘提取, 细胞神经网络, 阈值, 自适应