染发仿真及基于人脸检测的发型配准
Dye Hair Simulation and Hairstyle Registration Based on Face Detection
张璠璠1
作者信息
- 1. 四川大学计算机学院，成都 610065
- 折叠
摘要
提出将人脸检测用于发型自动配准以及发型染色的算法，通过灰度投影及Susan边缘检测和角点提取获得人脸的特征点，再和发型相对应的特征点匹配以达到发型配准；再通过三段折线式染色算法进行染色，既能使头发的颜色达到预期的目标，又能尽量保持原发型的亮度变化层次，达到有层次感的、自然的、不失真的染色效果。
Abstract
Proposes the face detection for the automatic registration of hair and hair dyed algorithm. Facial feature points obtained by the gray-level projection and Susan edge detection and corners match the corresponding feature points of hair to achieve the hairstyle registration. Through the dyeing algorithm of three sections of broken line type, can make the hair color to achieve the desired goal, and tries to keep the brightness of the original hairstyle change level to keep the administrative levels feeling, natural, undistorted dyeing effect.
关键词
人脸检测/灰度投影/角点检测/发型配准/染发仿真Key words
Face Detection/Gray-Level Projection/Corner Detection/Hair Registration/Dye Hair Simulation引用本文复制引用
出版年
2015
NETL
NSTL
维普
万方数据