基于形态学的二分法边缘提取算法

A bisectional edge extraction algorithm based on mathematical morphology

对于光照不均匀的图像,形态学边缘提取算法的分辨率远远逊色于人眼的分辨率.产生这种差异的原因是形态学算法仅仅是从几何学的角度出发来检测边缘,并没有模拟出人眼的生物特性.为了提高形态学算法的分辨率,通过研究人眼对光强的特性响应曲线,注意到了人眼对于光线具有亮度适应特性.把亮度适应特性加入形态学边缘提取算法,得到了高分辨率的二分法边缘提取算法.二分法边缘检测算法以强弱光亮度的中心点亮度为分界点,高于分界点的像素亮度被削弱,低于分界点的像素亮度被提高.如此在压制强光的同时增强弱光来模拟出人眼的亮度适应特性.实验证明二分法是一个具有高实时性、低噪声、高分辨率边缘提取算法.

The resolution of a morphological edge extraction algorithm is lower than that of the human eye for pictures with uneven illumination because the former only detects edges based on their angle of geometry rather than simulating the biological characteristics of the human eye. To improve the resolution of a morphological algorithm, the adaptability of the human eye to light luminance must be considered. Combining light luminance adaptability with a morphological edge extraction algorithm results in a new algorithm with high resolution. We have called it a bisectional edge extraction algorithm. It takes the luminance of a central point between a point with strong luminance and a point with weak luminance as a dividing point. Then pixels with luminance lower than that of the dividing point are suppressed, otherwise they are enhanced. In this way, the light adaptability of the human eye is simulated. Experiments show that this new algorithm is a good edge extraction algorithm with high resolution, real - time results and low noise.