【摘要】 提供了一种新的小波变换边缘检测方法，这种方法解决了第一代多尺度边缘检测方法中存在的两个问题即边缘仅定义为信号的奇异性表现；滤波尺度参数难以选择。它直接从人的视觉特征出发，描述边缘的特征，并定义象素点在邻域内的有效边缘测度，作为对该象素点进行滤波的尺度调整依据，以解决多尺度边缘检测中的尺度确定，从而实现边缘检测。实验和对比结果都证明，这种方法效果很好。更多还原

【Abstract】 ： A new edge detection method has been presented in this paper．It can solve the two problem that existed in the first generation wavelet transform actually，one is edge define only considering its oddity，another is to select filtering scale difficultly．The point of this method is from vision character，which depicts the edge nature and defines the efficient edge measure in the neighbor pixel filed．Based on these，it can adjust the filtering scale and solve the final scale of edge detection，so the detection can be realized．By many experiments and comparisons，the better effect can be testified and its use scope can be extended．更多还原