基于分形的低反差图像增强方法

An enhancement method for low contrast image based on fractal

对于工业现场采集的一类低反差、噪声大的图像,经过非线性滤波后根据分数布朗随机场模型理论,从灰度空间变换到分形参数空间,并以归一化后的像素分形维数为加权值对图像进行分形增强.非线性滤波刻意避开对图像边缘的平滑,能在消除图像噪声的同时很好地保持图像边缘等细节信息.与直方图的灰度拉伸增强比较,分形方法增强具有突显灰度差别不大的目标边缘等重要细节信息和抗噪能力强的优点,为后续的图像识别作了很好的准备工作.

In general, an image collected from industrial field is a kind of low contrast and noisy image. According to the theory of nonlinear fractional Brownian random field model, a new enhancement method is proposed in this paper, in which the image is transformed from gray space into fractal parameter space after the image is filtered through nonlinear filter. The normalized fractal dimension of each pixel in the image acting as weight will enhance the image. In contrast with the gray enhancement of histogram equalization, fractal enhancement method has the advantages in standing out the edges of gray target with no distinct difference and strong power in antinoise. It would be a good prepare＇for the image to be recognized in future.