Abstract: In the image-based medical diagnosis，effective image enhancement algorithm is required to dig decision-making information from image as much as possible.Because the wavelets fails to provide the optima representation of two-dimensional image with the “linear” or “face” singularity，on the basis of analyzing the human vision properties，a image enhancement algorithm based on the geometrical adaptive directional windows is presented.The wavelets domain is divided into the homogeneous windows and the structure geometrical windows by the use of geometrical characteristic of image.Firstly，the noise is restrained in the homogeneous windows；then the geometrical direction of the structure directional windows is computed，and 2-D information is projected on the 1-D along the certain directions，thereby 1-D signal is managed by wavelet transform for enhancement.The results of experiments show that the proposed algorithm not only can advance the contrast of image and improve the quality of image but also can resolve the problem of the noise magnified and extra-enhancement of clear edge，which is hard to process in the conventional algorithms.The handled image is advantageous for the doctor to carry on analysis diagnosis and follow- up processing of medical images.

Key words: image enhancement, direction windows, wavelet transform, medical diagnosis

摘要： 在影像的医学诊断中，为了从图像中挖掘出尽可能多的决策信息，需要有效的图像增强算法。在分析人眼视觉特性的基础上，针对小波不能“最优”表示含线、面奇异的二维图像的问题，提出了基于几何多尺度方向窗的图像增强算法。利用图像的几何特征，在小波域中划分出同质窗和结构方向窗。首先在同质窗内抑制噪声；然后计算出结构方向窗的几何方向，再沿窗方向作2-D 向1-D系数投影，进而对产生的1-D 信号做增强处理。实验结果表明，该方法在提高医学图像对比度、改善图像质量的同时有效地解决了传统算法中难以克服的噪声放大和清晰边缘过增强等问题，处理后的图像更利于医生进行分析诊断和医学影像的后续处理。

关键词: 图像增强, 几何窗, 小波变换, 医学诊断