摘要
针对传统图像清晰度评价算法在显微自动对焦系统中抗噪性弱的问题,在传统梯度函数基础上提出了一种改进的Sobel算子图像清晰度评价算法。首先利用自适应中值去噪算法对显微图像预处理,减少噪声干扰;其次增加Sobel算子模板方向,采用四方向模板来计算像素点梯度;最终使用自适应全局阈值确定图像目标边缘,使用强边缘像素点的梯度平方和作为清晰度评价值,在一定程度上有效地解决了传统评价算子方位局限、抗干扰性差等问题。实验结果表明,该算法与传统梯度评价函数相比,图像处理速度提升了15.6%,灵敏度因子平均提升了0.711 9,具有良好的单峰性、抗干扰性和灵敏度。
An improved Sobel operator image sharpness evaluation algorithm is proposed based on the traditional gradient function to solve the problem that the traditional image sharpness evaluation algorithm has weak noise resistance in the micro autofocus system.First,the adaptive median denoising algorithm is used to preprocess the micro-image to reduce the noise interference.Secondly,the Sobel operator template direction is increased,and the four-direction template is used to calculate the pixel gradient.Finally,an adaptive global threshold is used to determine the edge of the image target,and the sum of the gradient squares of the strong edge pixels is used as the clarity evaluation value,which effectively solves the problems such as the limitation of traditional evaluation operator orientation and poor anti-jamming.The experimental results show that the algorithm improves image processing speed by 15.6%and sensitivity factor by 0.7119 on average compared with the traditional gradient evaluation function.It has good unimodality,anti-jamming and sensitivity.
作者
张永超
李英
唐智勇
谢康旗
ZHANG Yongchao;LI Ying;TANG Zhiyong;XIE Kangqi(School of Opto-Electronic Engineering,Changchun University of Science and Technology,Changchun 130022)
出处
《长春理工大学学报（自然科学版）》
2023年第2期96-105,共10页
Journal of Changchun University of Science and Technology(Natural Science Edition)
基金
吉林省科技厅重点研发项目(20200401070GX)。
