摘要
介绍了一种皮肤图像症状的自动识别系统,利用马氏距离图像识别的算法,经过图像颜色空间转换、自适应法求取阈值进行图像分割、数学形态学后处理和边缘检测、症状区域的轮廓提取和种子填充等图像处理过程,最后提取了皮肤样本的26个特征参数,计算皮肤样本距离黄褐斑、粉刺、黑头、雀斑4种症状的马氏距离,根据马氏距离最小原则对被测皮肤样本进行症状归类。
An auto- recognition system for skin symptom image is introduced in the paper. This system utilizes Mahalanobis distance algorithm. Its image processing includes color space transformation, image segmentation according to threshold by self- adapting method, image post-processing by mathematical morphology, edge detection, contour tracing, seedfill and so on. Then, 26 characteristic parameters are extracted and the Mahalanobis distances between the new sample and chloasma, comedo, blackhead and ephelis are calculated separately. According to the minimum principle, the new samples are classified into different symptoms.
出处
《电视技术》
北大核心
2006年第1期86-89,共4页
Video Engineering
基金
上海市重点学科建设项目(项目编号:T0103)
关键词
马氏距离
图像分割
边缘检测
轮廓提取
种子填充
Mahalanobis distance
image segmentation
edge detection
contour tracing
seedfill
