DocumentCode :
259507
Title :
Clothing Extraction Using Region-Based Segmentation and Pixel-Level Refinement
Author :
Zhao-Rui Liu ; Xiao Wu ; Bo Zhao ; Qiang Peng
Author_Institution :
Dept. of Comput. Sci. & Eng., Southwest Jiaotong Univ., Chengdu, China
fYear :
2014
fDate :
10-12 Dec. 2014
Firstpage :
303
Lastpage :
310
Abstract :
In this paper, we demonstrate an effective method for automatic extracting clothing object from fashion photographs, an extremely challenging problem due to the non-uniform natural backgrounds, various types of apparel and different poses of human models. This method consists of three phases: (1) coarse clothing area localization by pose estimation and super pixel segmentation, (2) region-level image segmentation, (3) pixel-level refinement using spatial information and Grab cut. Experiments on a dataset with 1000 images crawled from Taobao demonstrate that the proposed method outperforms other methods, which can extract clothing from images with complex background.
Keywords :
clothing; feature extraction; graph theory; image segmentation; pose estimation; Grabcut; automatic clothing object extraction; coarse clothing area localization; fashion photographs; human model; nonuniform natural background; pose estimation; region level image segmentation; spatial information; superpixel segmentation; Clothing; Educational institutions; Estimation; Image color analysis; Image segmentation; Linear programming; Torso; clothing extraction; pixel-level refinement; region-based segmentation; superpixel;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Multimedia (ISM), 2014 IEEE International Symposium on
Conference_Location :
Taichung
Print_ISBN :
978-1-4799-4312-8
Type :
conf
DOI :
10.1109/ISM.2014.74
Filename :
7033043
Link To Document :
بازگشت