使用直方图熵阈值法,从背景中自动分离出医学心脏图象

在医学科学研究和临床诊断中,常常需要在图象背景中分离出心脏或其他内脏器官的图象,以便对心脏或内脏器官进行几何学参数的测量。本文使用信息论中熵概念,推导出与先验和后验信息有关的评价函数,由它来选择最佳阈值,使分离出完整的心脏图象。为了能得到精确、清晰的心脏边界,本文使用了双曲线函数增强处理,选择常数C为0.5或1。采用灰度值窗口法后,并把阈值处理后图象的直方图起点移至灰度级为零的位置,使增强心脏边界的效果更为明显。本文使用Roberts边缘检测法获得了精确又清晰的心脏边界。

Auto-seperated Cardiac Image with the Method of Histogram Entropy Threshold

The image of heart or other internal organ has to be seperated from the background for measuring geometric parameters in medical science and clinical examination. The conception of entropy of information theory is used to deduce an evaluation function relating to the prior and posterior information. An integrate heart image can be extracted from its background after the optimum threshold value is chosen. In order to obtain the exact and clear image edge of a heart,the enhancement with hyperbolic function is used in which the constant C is 0.5 or 1. By the adoption of gray level window, the effect fo enhancement of heart boundary is more obvious if the original point of histogram of the image which has been delt with the optimum threshold, is moved to the position of O level.An exact and clear heart boundary can be obtained with the Roberts dge edetection algorithm.