基于结构保持因子的显微图像滤波质量客观评价指标

【摘要】目的:解决计算机辅助精子质量分析系统缺乏显微图像滤波算法性能客观评价指标的问题。方法:首先对精子跟踪轨迹连续性、滤波图像直方图、滤波算法性能三者之间的相关性进行研究，然后对不同噪声密度下的相关性进行量化计算，最后建立结构保持因子数学计算模型。结果:滤波算法的结构保持因子越大，精子跟踪轨迹连续性等级越高，证明结构保持因子作为显微图像滤波性能衡量的有效性。结论:结构保持因子克服了显微图像滤波算法性能评价的主观性，该指标可为精子显微图像滤波算法的选取提供理论指导，从计算机图像处理角度为计算机精子质量检测系统提供质控依据。

An indicator for objectively evaluating the quality of microscopic image filtering based on structure preservation factor

Abstract: Objective To solve the problem that computer-aided sperm analysis system lacks of objective indicators for evaluating the performance of microscopic image filtering algorithms. Methods The correlation among continuity of sperm trajectory tracking, gray-level histogram of filtering image and performance of filtering algorithm was analyzed and then, the correlation under different noise densities was quantitatively calculated. Finally, a mathematical model of structure preservation factor was established. Results The larger the structure preservation factor of the filtering algorithm was, the higher the continuity level of sperm trajectory tracking was, which proved the effectiveness of the structure preservation factor as a measurement of filtering performance. Conclusion The structure preservation factor overcomes the subjectivity of the performance evaluation of filtering algorithm for microscopic image. The proposed indicator can not only provide theoretical guidance for the selection of filtering algorithm for sperm microscopic image, but also provide quality control basis for computer-aided sperm analysis system from the perspective of computer image processing.