免散瞳眼底照相机自动对焦算法设计

Auto-focusing Algorithm Design for Non-mydriatic Fundus Camera

眼底图像的自动对焦是眼底照相机的重要组成部分。在分析各种应用于眼底图像自动对焦算法的基础上,针对免散瞳眼底照相机及眼底图像对比度特点,在现有低对比度图像自动对焦算法的基础上,提出了基于区域分块的自动对焦改进算法。算法结合无重叠型的采样方式与梯度向量平方函数,在保证对焦灵敏度的基础上,提升了自动对焦的运算速度。仿真实验证明,该算法与原算法及常用自动对焦算法相比具有较好的灵敏度及运算速度,能够应用于实际的眼底图像自动对焦中。

Auto-focusing algorithm is critical for the fundus camera. Based on the analysis of different auto-focusing algorithms for retinal image and the characteristics of non-mydriatic fundus camera and retinal image contrast, an improved auto-focusing algorithm based regional scales is presented. The algorithm combines non-overlapping subsampling with gradient vector square function, improving auto-focusing speed while conserving sensitivity. Experimental results show that the new method has higher sensitivity and speed than the orignal algorithm and other auto-focusing algorithms available, andmeets the requirement of the retinal image auto-focusing.