全方向M型心动图中带Contourlet变换的检测算法

摘要

为了提高全方向M型心动图运动曲线检测效果,根据多尺度分析的思想,提出一种基于Contourlet变换的边缘检测算法.首先利用基于局部熵的算法对图像进行检测尺度的划分,然后在单一尺度空间上借鉴小波模极大值检测思想对图像边缘进行提取,最后利用补偿连接方法融合不同尺度空间下的边缘曲线,以得到最佳的运动曲线.实验结果表明,该算法可以对心动图运动曲线从粗分辨率到细分辨率进行连续逼近,检测效果良好.%In order to improve the detection effect of omni-directional M-mode echocardiography motion curve,the detection algorithm based on Contourlet transform is proposed according to the idea of multi-scale analysis.Firstly it uses the local entropy-based algorithm to divide the image into images at different scales.Secondly the edges of the images are extracted in the space of a single scale with the help of the wavelet transform modulus maxima method.Finally it integrates the edge curves under different scale spaces with a compensation link method to get the best motion edges.Experimental results show that the algorithm can continuously approximate echocardiography motion curves from coarse to fine resolutions and obtain good results.