自動閥值選取是彩色圖像邊緣檢測的難點和關鍵問題，首先從人類的視覺特性出發，分析了人類視覺系統的亮度感知門限函數，利用數學建模方法和色度學方面的研究成果，給出人類視覺系統對彩色圖像邊緣所能識別的最小彩色差函數。利用小波邊緣檢測原理，給出一種能根據不同彩色背景自動選取閥值的小波邊緣檢測算法. 仿真實驗證明算法有較好的自適應性和良好的邊緣識別性能。Automatic threshold choice is difficult and important to color images edge detection, First of all, on the base of human visual properties, threshold versus intensity function of human visual system are researched and analyzed, by using method of mathematic modeling and the result of saturation research, function of least color difference that can be perceptible by visual systems in detecting edge is given. Finally, we use the principle of wavelet edge detection and suggest an algorithm of automatic threshold choice according to difference color background in wavelet edge detection. Emulate experiments demonstrate good adaptability and edge recognition effeteness of the algorithm.