用分形理论增强红外图像的方法  

Enhancement of infrared image using fractal theory

作　　者：于天河[1,2] 戴景民[1] 贾丽娟[2] 

机构地区：[1]哈尔滨工业大学电气工程及自动化学院,哈尔滨150001 [2]哈尔滨理工大学计算中心,哈尔滨150080

出　　处：《哈尔滨工业大学学报》2009年第11期77-80,共4页Journal of Harbin Institute of Technology

基　　金：黑龙江省自然科学基金资助项目(F200818)

摘　　要：红外图像边缘模糊不利于人眼观察,为了改善红外图像的视觉效果,提高红外图像的可视度,采用分形布朗理论对红外图像进行增强.由分形布朗随机场的性质可知,图像的小区域范围满足自相似性,但区域的边界处的规律性被打破,因此边界的H值发生奇异,红外图像灰度表面粗糙度即被描述出来.首先计算图像每个像素的分形参数,根据人眼的视觉敏感特征把图像的像素进行分类,分为平滑点和边缘点,然后对各个像素分别加权增强.试验结果表明,增强图像突出了目标的轮廓,获得良好的视觉效果.由于该方法充分考虑视觉特性,可以解决红外图像边缘模糊可视性差的问题.In order to improve the visibility of infrared image, the fractal Brownian theory is employed to enhance the visual effect of infrared images. According to the characteristc of fractal Brownian random （FBR） fields, small regions in the image are self-similar in statistics, while the regularity at the edge of objects in the image is broken, therefore H value of the edge is in singularity. The fractal dimension of each pixel is calculated using a FBR model, then the self-similitude and the surface roughness of the gray level of each pixel are described using the fractal value of each pixel. Human visual system （HVS） is sensitive to the edge area of an image, and the pixels are then classified into edge pixels and smooth pixels according to their fractal dimen- sions. The gray level of each edge pixel is weighted and enhanced using a fractal based HVS. Experimental results indicate that FBR model is effective in detecting the edge detail of an infrared image. This approach can be used to preserve edge details and enhance the visibility of an image.

关 键 词：红外图像 分形 视觉特性 图像增强 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]