基于形态学的角膜地形图特征提取方法研究

The Study of Corneal Topography Feature Extraction Method Based on Morphology

提出了一种新的角膜地形图像特征提取算法。首先利用区域图像计算偏心率,用自适应确定同心圆的圆心,以此为极点将图像转至极坐标域,利用形态学的小尺度方形结构算子及图像的黑TOP-HAT变换补得到去除噪声后的滤波图像;其次利用给定方向上相距固定距离的点对腐蚀图像,按照方向强度确定出纹理直线的主要方向,利用开运算提取有效的线形特征;最后利用闭运算,对因内睫毛影响而损失的数据进行有效弥补,获得圆环边缘检测完整的结果图像。经多幅实验数据证明,该算法满足后续参数精确计算的要求,具有数据损失小,数据精度高,算法易实现的特点。

This paper puts forward a new retrieval algorithm of corneal topography features based on morphology. Firstly, eccentricity is calculated by use of regional image to self-adapt and locate the center of concentric circles, with which as polar point the image is transferred into polar coordinates, where morphological small scale square structure operator and black TOP-HAT Transformation are used to get rid of noises and get filtered image. Secondly, corrosion images of point pairs with fixed distance in a given direction are used to figure out main directions of fabric texture lines according to direction strength, in which effective linear features are extracted by opening operation. Finally, closing operation is used to make up for the lost data caused by inner eyelashes, obtaining a complete image for ring edge detection. Experiments show that the algo- rithms meet the needs of precisely calculating parameters, with the characteristics of less data loss, higher data accuracy and easier availability.