摘要
为了克服因静脉图像照度不均造成的现有分割算法对静脉纹路分割不够精确的问题,提出了一种基于最大邻域内差(Maximal intra-neighbor difference,MIND)的静脉图像分割算法,其核心是充分利用静脉图像的邻域信息和新设计的距离函数计算出原图的MIND图像,并与经过直方图修正后的原图加权相加得到了增强图像,之后,通过计算出增强图像的均值图像并与增强图像进行加权比较得到最终的分割结果。在分割的过程中,可以根据MIND图像的直方图自适应调整算法中的分割参数提高分割效果,最后的实验结果证明了算法的有效性。
In order to overcome the problems that vein lines are inaccurate in results of existed segmentation algorithms caused by the uneven illumination of the vein image, a novel image segmentation algorithm based on maximalson intra-neighbor difference（MIND） is proposed. In the algorithm, the neighbor information of original vein image is well used to get the MIND image together with a new distance function. The enhanced image is got by adding with original image processed by histogram modification. Through the calculated mean image of enhanced image and weighted comparison with enhanced image, segmentation result is obtained. During the segmentation, to the segmentation result can be improved by adjusting key parameters adaptively according to the histogram of MIND image. Experimental results demonstrate the effectiveness of the proposed algorithm.
出处
《光学学报》
EI
CAS
CSCD
北大核心
2009年第7期1830-1837,共8页
Acta Optica Sinica
基金
国家自然科学基金(60674034)
佛山市科技发展专项资金(200601006)
佛山市禅城区产学研专项资金 (2007B1041)资助课题
关键词
图像处理
静脉识别
图像分割和增强
最大邻域内差
直方图
image processing
vein recognition
image segmentation and enhancement
maximal intra-neighbor difference （MIND）
histogram
