摘要
自动聚焦是免散瞳眼底照相机的一项关键技术,在分析眼底照相机成像原理及眼底图像特点的基础上,提出了一种应用于免散瞳眼底照相机的粗精结合自动聚焦方法。提出结合眼底图像视场特点的内接正方形固定聚焦窗口选择方法,并结合Robert梯度函数实现粗聚焦。采用基于局部信息熵的自动聚焦窗口选择算法得到精聚焦中的聚焦窗口,并提出基于9pixel×9pixel尺寸泽尼克正交矩的图像清晰度评价函数实现精聚焦。实验结果表明该方法中基于内接正方形的固定聚焦窗口选择方法具有更好的稳定性,精聚焦中的图像清晰度评价函数具有更高的灵敏度与实时性。
Auto-focusing is one of the key issues in the non-mydriatic fundus camera, a fine and rough combined auto-focusing method is presented after analyzing imaging principle of the fundus camera and the retinal image characteristics. Inscribed square fixed focusing window selection method combined with characteristics of fundus image field is presented, and Robert gradient function is used to realize the rough focusing. The focusing window for fine focusing is selected through auto-focusing window selection algorithm based on local entropy, and image sharpness evaluation function based on Zernike orthogonal moment with 9 pixel× 9 pixel mask is presented to realize the fine focusing. The experimental results show that the focusing window method based on inscribed square is more stable, and proposed image sharpness evaluation function for fine focusing has better sensitiveness and realtime performance.
出处
《光学学报》
EI
CAS
CSCD
北大核心
2014年第11期109-120,共12页
Acta Optica Sinica
基金
国家自然科学基金(60827002)
福建省教育厅教育科研项目(JB13148)
关键词
图像处理
自动聚焦
聚焦窗口
内接正方形
泽尼克正交矩
image processing
auto-focusing
focusing window
inscribed square
Zernike orthogonal moments
