一种基于小波变换的数字图像自动聚焦算法

由傅立叶光学理论可知 ,图像高频能量值可以反映聚焦程度 .在传统图像灰度方差评价函数的基础上 ,利用小波多分辨率分析 ,提出了一种新的自动聚焦算法 .清晰度评价函数通过舍去图像低频概貌信号 ,突出了高频能量 ,具有更好的尖锐性 .并分析了分解层数对评价函数性能的影响 .通过改进的变步长爬山算法实验表明 ,该算法与传统灰度方差算法相比具有更好的聚焦精确度 .

Auto-focusing algorithms of digital image based on wavelet transform

According to Fourier optical theory, the high frequent energy of image can expresse the focus condition. On the basis of evaluation function of using image intensity variance, we proposed a new auto-focusing algorithm by means of Wavelet Transform Multi-Resolution Analysis(WTMRA) theory. The proposed evaluation function gained sharper performance because it abandoned the sub-image of low frequent signal and strengthened the high frequent energy. The MRA decomposed layers influence on evaluation performance was also referred in this paper. Comparing with the existing algorithms, the experiments adopting alter-scale Mountain Climbing Servo showed that the proposed algorithm has a characteristic of precision in auto-focusing.