昆虫翅膀运动变形测量中的图像特征提取  

Image feature extraction in insect kinetic deformation measurement

作　　者：江洁[1] 费金娥[1] 王颖[1] 张广军[1] 

机构地区：[1]北京航空航天大学仪器科学与光电工程学院,北京100083

出　　处：《光学精密工程》2007年第10期1616-1621,共6页Optics and Precision Engineering

基　　金：国家杰出青年科学基金资助项目(No.50125518)

摘　　要：通过分析翅脉图像的特点,提出了一种用于提取昆虫图像翅脉和翅脉交点的亚像素级图像处理方法。该方法应用小波理论对图像消噪,并采用中值滤波预处理图像,然后利用小波同态滤波和双线性广义模糊相结合的算法增强图像的翅脉信号。应用基于Hession矩阵的亚像素提取算法提取翅脉中心点的位置,并根据翅脉方向变化的连续性,利用邻域角度约束方法确定翅脉交叉点。最后,给出几种算法实验结果的比较和分析,结果表明,该方法成功实现了翅脉和翅脉交点特征的提取,并有效抑制了噪声和干扰。An image processing method for extracting sub-pixel of wing veins and vein junctions of insect images is proposed. The image of wing veins is denoised by wavelet, and then enhanced by wavelet homomorphic filter and bilinear generalized fuzzy. The center positions of wing veins are extracted by the sub-pixel extracting algorithm based on Hession matrix, and vein junctions are determined by the neighboring angle constraint and continuity of vein direction. The experiment results show that compared with other methods, the proposed approach can extract wing veins and vein junctions successfully, and can suppress the noise and disturbance effectively.

关 键 词：图像处理 翅脉 小波同态滤波 广义模糊增强 特征提取 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]