利用图像处理方法进行自动调焦的关键是提取图像清晰度特征,并建立其评价算法。本文研究了灰度值线性变换、灰度直方图均衡、中值滤波及同态滤波等预处理方法和基于功率谱的清晰度评价函数,并与其它的评价方法进行了比较分析。研究表明,中值滤波和灰度值线性变换相结合的预处理方法,具有效果好、计算量少等优点;基于功率谱的清晰度评价函数比其它的评价方法具有更好的调焦性能和更明确的物理意义。根据基于功率谱的图像调焦算法的特点,设计了图像处理模块的结构框架和算法流程,提出了流水线作业结构、“乒乓”操作模式、双蝶形处理器复用、基-2FFT算法的FPGA实现方案,提高了图像自动调焦的计算和响应速度。
The key to focus automatically using the image processing method is to take the character of image definition and to establish its evaluation algorithm. The paper studied the preprocessing methods,, such as the linear transform of gray value, the gray histogram equilibrium, the median filter, the homomorphic filter, and the image definition evaluation function based on power spectrum. The evaluation method was analyzed and compared with other methods. The results indicate that the preprocessing method of combining the median filter with the linear transform of gray value is good in effect, little in calculation and so on. The image definition evaluation function based on power spectrum is superior in focusing performance and more definite in physics meaning to other methods. According to the characteristic of the image focusing algorithm based on power spectrum, the structure frame and algorithm flow of the image processing module were designed, and the implementation solution of the pipelining structure, the 'ping-pong' working mode, the sharing processor of double butterfly shape and the base-2 FFT algorithm in FPGA were presented in this paper. The above-mentioned measures can improve the computing and response speed of the image auto-focus.