DocumentCode :
3289067
Title :
Face Detection Method Based on a New Nonlinear Transformation of Color Spaces
Author :
Zhang, Zhengzhen ; Shi, Yuexiang
Author_Institution :
Sch. of Inf. Eng., Xingtan Univ., Xiangtan
Volume :
4
fYear :
2008
fDate :
18-20 Oct. 2008
Firstpage :
34
Lastpage :
38
Abstract :
In this paper, against Hsu R L skin tone detection algorithm, which has high missing detection ratio when the facial brightness of an image is high, we introduce a new nonlinear transformation method of the YCb "Cr" color skin space, which is the color segmentation of the human face for regional analysis and extraction; Then implement image rotation and template matching. Experimental results indicate that when the facial brightness of an image is high, the new proposed nonlinear color space conversion algorithm based on YCbCr color space is better than the simple method, which wipes off the effect of the brightness directly, and Hsu R L skin detection algorithm. The series of experimental results demonstrate clearly that our new method has a higher detection ratio on high bright facial images, and can drop the false or missing detection ratio on high luminance images. Besides, we have more detailed testing results on the eyes, eyebrows, mouth and other parts.
Keywords :
face recognition; image colour analysis; image matching; image segmentation; Hsu RL skin tone detection algorithm; YCbCr color skin space; color segmentation; face detection method; human face; image rotation; image template matching; nonlinear color space conversion algorithm; nonlinear transformation; regional analysis; Algorithm design and analysis; Brightness; Chromium; Detection algorithms; Face detection; Humans; Image analysis; Image color analysis; Image segmentation; Skin; Face detection; Gauss model; Image segmentation; YCgCr color space;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Fuzzy Systems and Knowledge Discovery, 2008. FSKD '08. Fifth International Conference on
Conference_Location :
Jinan Shandong
Print_ISBN :
978-0-7695-3305-6
Type :
conf
DOI :
10.1109/FSKD.2008.583
Filename :
4666351
Link To Document :
بازگشت