(JBE Vol. 19, No. 3, May 2014) (Regular Paper) 19 3, 2014 5 (JBE Vol. 19, No. 3, May 2014) http//dx.doi.org/10.5909/jbe.2014.19.3.396 ISSN 2287-9137 (Online) ISSN 1226-7953 (Print) a), a), a), a) Invariant Classification and Detection for Cloth Searching Inseong Hwang a), Beobkeun Cho a), Seungwoo Jeon a), and Yunsik Choe a).,,,., LBPROT_35. LBP_ROT(Local Binary Pattern with ROTation-invariant),. 11,., 810 36, 94.4% Dense-SIFT. Abstract The field of searching clothing, which is very difficult due to the nature of the informal sector, has been in an effort to reduce the recognition error and computational complexity. However, there is no concrete examples of the whole progress of learning and recognizing for cloth, and the related technologies are still showing many limitations. In this paper, the whole process including identifying both the person and cloth in an image and analyzing both its color and texture pattern is specifically shown for classification. Especially, deformable search descriptor, LBPROT_35 is proposed for identifying the pattern of clothing. The proposed method is scale and rotation invariant, so we can obtain even higher detection rate even though the scale and angle of the image changes. In addition, the color classifier with the color space quantization is proposed not to loose color similarity. In simulation, we build database by training a total of 810 images from the clothing images on the internet, and test some of them. As a result, the proposed method shows a good performance as it has 94.4% matching rate while the former Dense-SIFT method has 63.9%. Keyword informal, cloth, pattern, descriptor, classification a) (Yonsei University Electrical and Electronic Eng.) Corresponding Author (Inseong Hwang) E-mail drpro@nate.com Tel +82-2-2123-2774 " IT /IT". (NIPA- 2014-H0301-14-1012) Manuscript received March 19, 2014 Revised April 21, 2014 Accepted April 21, 2014
3 (Inseong Hwang et al. Invariant Classification and Detection for Cloth Searching)..,,.,.,,, [1][2][3][4][5].,,,,., LBPROT_ 35. Dense-SIFT, LBP_ROT(Local Binary Pattern with ROTation-invariant) [6],. HLS 11,., Dense-SIFT [8][9]. Dense- SIFT SIFT Dense-SIFT LBPROT_35 HLS., 810.,,.,,... 1. LBPROT LBP (Local Binary Pattern) 1, 5 1, 0 1. LBP (Local Binary Pattern) Fig. 1. LBP (Local Binary Pattern) calculation process
(JBE Vol. 19, No. 3, May 2014), () 2, LBP 53 [10][11][12][13]. LBP 0 255 256, 36, LBP_ROT (Local Binary Pattern with ROTation-invariant) [6]. LBP_ROT LBP., 2 LBP_ROT = 53 LBP_ROT = 106 45. LBP. 2. LBP 45 LBP Fig. 2. Comparing between LBP and 45 rotated LBP LBP_ROT. LBP_ROT., LBP_ROT /.,,. 1, 2, 2, 2, 4., 4 2 LBP_ROT [6], 2 4., LBP_ROT,, LBP_ROT., 3 90 4. 2 LBP_ROT Fig. 4. LBP_ROT histogram comparing for original image and magnifying patch image by 2 times 3. 90 LBP_ROT Fig. 3. LBP_ROT histogram comparing for original and 45 rotated patch image LBP_ROT.
3 (Inseong Hwang et al. Invariant Classification and Detection for Cloth Searching) [14]., LBP_ROT LBP_ROT, 36 LBP_ROT LBP_ROT LBPROT_35. 5. 1.5 ( 1), 2 ( 2) LBPROT_35 Fig. 5. LBP_ROT_35 histogram comparing for original image and magnifying patch image by 1.5 and 2 times LBPROT_35 5. 2. HLS.,..,,.,.. RGB HLS [15][16]. HLS RGB (Hue), (Saturation), (Lightness). (Hue, H) (Color Circle) 0. H 0 ~360, 360 0, (Saturation, S) 100%, 0%, (Lightness, L),. RGB HLS., (L) (H) (S). HLS,. (H) (S), (L) 15%, 80%. 15%~80%,,,,,,,,., (L) (S) 10%, (15%<L<80%) (10%<S) (H) 344 ~360, 0 ~9, 10 ~37, 38 ~65, 66 ~145, 146 ~181, 182 ~255, 256 ~315, 316 ~344. 11,. 11.
(JBE Vol. 19, No. 3, May 2014) 6 11. 11,. 6. 11 Fig. 6. Two examples of HLS-based color histogram Histogram Intersection, Dense-SIFT LBPROT_35 Histogram Intersection [17].. 1. 7, Dense-SIFT [8][9] LBPROT_35 HLS,., Dense-SIFT, 128 SIFT, K-means, 7. Fig. 7. Block diagram of the whole process K-D [18][19] bin. LBPROT_35, LBPROT_35 35 bin. 11 bin.,,
3 (Inseong Hwang et al. Invariant Classification and Detection for Cloth Searching),. 2.. Dense-SIFT,, Dense-SIFT LBPROT_35. 7,,,., Dense-SIFT [8][9], K-D, LBPROT_35,, Histogram Intersection [17],.. 810 100x100,. 36 10,., Dense-SIFT 32x32 4 128, 4, 100x100 361. 361 128 SIFT, 810 361x810 SIFT K K-means [20][21], 300., K-means K-D [18][19], 8. Dense-SIFT ( 5 10 ) Fig. 8. Comparing between the proposed method and Dense-SIFT(in case of Top5 and Top10), 8, 36, 5, Dense-SIFT, 10, 29 Dense-SIFT.,,, Dense-SIFT 36 23 (63.9%), 34 (94.4%). 9. Dense-SIFT. LBPROT_35 feature HLS
(JBE Vol. 19, No. 3, May 2014). 10. Dense-SIFT ( ) Fig. 10. comparing the result of scaled image between the proposed and Dense-SIFT 9. Dense-SIFT ( ) Fig. 9. comparing the result images between the proposed and Dense-SIFT 9, 29, Dense-SIFT 10,, 1, Dense- SIFT, 5 1., 10 (50 ~ 200%). Dense-SIFT,, Top5 Top10 3, Top5, Top10. LBPROT_35. 1. PC (Intel Xeon CPU 2GHz, 16GB RAM), Dense-SIFT LBPROT_35 HLS 11.5%, 6.9%,, 0.31, 0.29. 1. Table 1. Time consumption comparison between the proposed method and Dense-SIFT (sec) Dense-SIFT 0.29/1 0.26/1 0.31 Dense-SIFT 0.29. 11.5% 6.9%. SIFT,
3 (Inseong Hwang et al. Invariant Classification and Detection for Cloth Searching), LBPROT_35 HLS. LBPROT_35, LBP_ROT,,, HLS 11 11,. Dense-SIFT, 94.4% Dense-SIFT 63.9%, 5 Dense-SIFT., 10. Dense-SIFT. (References) [1] Kumar, S. Suresh, and L. Ganesan. "Texture classification using wavelet based laws energy measure." International Journal of Soft Computing 3.4, 293-296, 2008. [2] Yang, Ming, and Kai Yu. "Real-time clothing recognition in surveillance videos." Image Processing (ICIP), 2011 18th IEEE International Conference on. IEEE, 2011. [3] Zhang, Jianguo, et al. "Local features and kernels for classification of texture and object categories A comprehensive study." International journal of computer vision 73.2 (2007) 213-238, 2007. [4] Pietikäinen, Matti, Timo Ojala, and Zelin Xu. "Rotation-invariant texture classification using feature distributions." Pattern Recognition 33.1 (2000) 43-52, 2000. [5] Yuan, Shuai, YingLi Tian, and Aries Arditi. "Clothing matching for visually impaired persons." Technology and disability 23.2 (2011) 75-85, 2011. [6] Timo Ahonen, Jiri Matas, Chu He, and Matti Pietikanen, Rotation Invariant Image Description with Local Binary Pattern Histogram Fourier Features, LNCS 5575, pp. 61-70, 2009. [7] Viola, Paul, and Michael J. Jones. "Robust real-time face detection." International journal of computer vision 57.2 (2004) 137-154, 2004. [8] D. G. Lowe. Object recognition from local scale-invariant features. In IEEE International Conference on Computer Vision (ICCV), pages 1150 1157, Kerkyra, Greece, 1999. [9] C. Liu, J. Yuen, A. Torralba, J. Sivic, and W. T. Freeman. SIFT flowdense correspondence across different scenes. In European Conference on Computer Vision (ECCV), 2008. [10] L. Wolf, T. Hassner, and Y. Taigman, Descriptor based methods in the wild, in Proc. ECCV, 2008. [11] J. Ruiz-del-Solar, R. Verschae, and M. Correa, Recognition of faces in unconstrained environments A comparative study, EURASIP Journal on Advances in Signal Processing, vol. 2009, pp. 1 20, 2009. [12] T. Ahonen, A. Hadid, and M. Pietikainen, Face description with local binary patterns Application to face recognition, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 28, no. 12, pp. 2037 2041, 2006. [13] Y. Rodriguez and S. Marcel, Face authentication using adapted local binary pattern histograms, Lecture Notes in Computer Science, vol. 3954, p. 321, 2006. [14] Canny, John. "A computational approach to edge detection." Pattern Analysis and Machine Intelligence, IEEE Transactions on 6 (1986) 679-698, 1986. [15] Joblove, George H. and Greenberg, Donald (August 1978). "Color spaces for computer graphics". Computer Graphics 12 (3) 20 25, 1978. [16] Swain, Michael J., and Dana H. Ballard. "Color indexing." International journal of computer vision 7.1 (1991) 11-32, 1991. [17] M. J. Swain and D. H. Ballard, Color indexing, IJCV, vol.7, no. 1, pp. 11 32, 1991. [18] J. L. Bentley. Multidimensional binary search trees used for associative searching. Communications of the ACM, 18(9)509-517, 1975. [19] Silpa-Anan, Chanop, and Richard Hartley. "Optimised KD-trees for fast image descriptor matching." Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference on. IEEE, 2008. [20] Xindong Wu, Vipin Kumar, J. Ross Quinlan, Joydeep Ghosh, Qiang Yang, Hiroshi Motoda, Geoffrey J. McLachlan, Angus Ng, Bing Liu, Philip S. Yu, Zhi-Hua Zhou, Michael Steinbach, David J. Hand, and Dan Steinberg. 2007. Top 10 algorithms in data mining. Knowl. Inf. Syst. vol.14, no.1, pp.1-37, Dec. 2007. [21] Elkan, Charles. "Using the triangle inequality to accelerate k-means." ICML. Vol. 3. 2003.
404 방송공학회논문지 제19권 제3호, 2014년 5월 (JBE Vol. 19, No. 3, May 2014) 저자소개 황인성 년 인하대학교 전자공학과 석사 현재 연세대학교 전기전자공학부 박사과정, SK플래닛 근무 주관심분야 영상 검색 및 영상 신호처리, 비디오 압축 등 - 1995 - 조법근 년 월 연세대학교 전기전자공학부 학사 현재 연세대학교 전기전자공학부 석사과정 주관심분야 패턴인식, 기계학습 등 - 2013 2 전승우 년 월 연세대학교 전기전자공학부 학사 현재 연세대학교 전기전자공학부 석박통합과정 주관심분야 패턴인식, 기계학습 등 - 2014 2 최윤식 - 년 연세대학교 전기공학과 학사 년 클리브랜드 케이스 웨스턴 리저브 대학교 석사 년 팬실베니아 주립 대학교 석사 년 일리노이즈 퍼듀 대학교 박사 년 ~ 1993년 현대전자 산업전자연구소 수석연구원 년 ~ 현재 연세대학교 전기전자공학부 교수 주관심분야 비디오 코딩, 비디오 통신, 통계적 신호처리, 디지털영상처리 등 1979 1984 1987 1990 1990 1993