Abstract: Detection and extraction of facial spots, wrinkles and pores for diagnosis of dermatosis, decision of efficacy of cosmetic and evaluation of skin are necessary. It selects a rectangular area in the facial skin image, diagnoses this area by using a filter bank which includes a wavelet homomorphic filter for detecting the spots and a ε-filter bank for detecting the wrinkles and pores. It extracts them by combining their characteristics with some morphological algorithms in image processing. It validates the accuracy of the method on the analysis of experimental result.

Key words: image diagnosis, wavelet homomorphic filtering, ε-filer bank

摘要： 在皮肤病诊断、化妆品功效判定和皮肤评价中都需要精确地检测和提取色斑、皱纹和毛孔。在人脸皮肤图像中截取一块矩形区域，设计了一个滤波器组对该区域图像进行诊断。该滤波器组包括一个小波同态滤波器和一个ε-滤波器组。一方面，把区域图像用小波同态滤波器滤波，进而检测出色斑。另一方面，把区域图像用ε-滤波器组滤波，进而检测出皱纹和毛孔。结合它们各自的特点和一些形态学图像处理算法分别对它们进行了提取。根据实验结果进行了分析，验证了方法的准确性。

关键词: 图像诊断, 小波同态滤波, &epsilon, -滤波器组