图像增强算法与评价方法研究  

Research on Image Enhancement Algorithm and Evaluation Method

作　　者：孙蕾[1] 温有奎[1] 李丙春[2] 

机构地区：[1]西安电子科技大学经济管理学院,西安710071 [2]喀什师范学院网络中心,喀什844000

出　　处：《计算机工程》2007年第8期168-169,172,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60372072)

摘　　要：研究了适应于乳腺钼靶X线图像特点的图像增强算法,探索量化评价增强算法效果的指标和体系。应用边界、标准偏差、熵以及4阶矩定义灰度均匀度,提出了增强放大因子来调节对比度增强程度的策略,同时给出了基于均匀度实现图像增强的算法,提出了4个增强效果量化评价指标。并与已实现的数学形态学增强算法和直方图均衡化算法进行了量化对比分析。综合分析指标表明,该算法优于数学形态学和直方图均衡化算法,但需要大量的实验和对比分析,研究了DSM、TBCs、TBCe和综合指标之间的相关性,从而找到了更适合于乳腺钼靶X线图像的对比度增强算法,探索出了更加科学的评价体系和指标。The paper studies enhancement algorithm on mammograms, and quantitative measures of contrast enhancement. The gray homogeneity is defined by edge value, standard deviation, entropy and the fourth moment for mammograms. The amplifier is presented to adjust the contrast enhancement in the whole image. Then the homogeneity based-on algorithm is applied to enhance mammograms. Four quantitative measures of contrast enhancement are put forward. The presented algorithm is compared with two other existing contrast enhancement techniques-histogram equalization and morphological enhancement. The index shows that the presented algorithm performs better than histogram equalization and morphological enhancement. A lot of experiments and comparisons are needed to study the relativity among DSM, TBCs, TBCe and the final index. Based on those, contrast enhancement algorithm for mammograms and rational evaluation system and measures can be found.

关 键 词：图像增强 灰度均匀度 直方图均衡化 评价 

分 类 号：TN919.81[电子电信—通信与信息系统]