22 A person recognition using color information 1110372 2011 2 13
,,.,.,,.,.,.,.,,.,..,,,, i
Abstract A person recognition using color information Tatsumo HOJI Recently, for the purpose of collection of marketing information, the technique of analyzing a person s action using the surveillance camera at a store is proposed. In order to analyze the person action in a store, it is necessary to continue pursuing a specific person. However, because of the complexity of human behavior, people can not recognize correctly. Therefore, the person s pursuit is interrupted and an accurate action analysis is difficult. Then The recognition that is stubborn for the change of the person is necessary. The purpose of the present study is to do stubborn recognition to the person s change by using information that can be acquired from one camera. Previous studies have focused on terms of contextual information for person recognition. It recognizes it by using person s color information. In the proposal technique, it corresponds to the change in person s direction by using color information on each head and body with person s impression color. The validity of the method proposed from experimental results. key words Person recognition, Color information, Context information, Surveillance camera, Action analysis ii
1 1 1.1.................... 1 1.2........................ 2 1.3................................... 2 1.4................................. 2 2 3 2.1............................. 3 2.1.1.......................... 3 2.1.2......................... 4 2.1.3...................... 4 2.2.......................... 4 2.2.1 MRF.............................. 5 2.2.2................... 5 3 6 3.1................................ 6 3.2.................................. 6 3.3.................................. 7 3.4................................. 8 3.4.1............................. 8 3.4.2............................. 8 3.5............................ 11 3.5.1...................... 11 iii
3.5.2............................... 12 3.5.3............................. 12 3.5.4............................... 12 4 14 4.1................................... 14 4.2................................. 15 4.3................................... 15 4.4...................................... 15 5 17 18 19 iv
2.1................................. 4 2.2............................ 5 2.3............................ 5 3.1................................ 7 3.2.............................. 8 3.3................................. 9 3.4 R................................... 9 3.5 G................................... 10 3.6 B................................... 10 3.7............................. 11 3.8................................. 12 4.1................................ 14 v
3.1,, (, )............. 13 4.1.................................... 15 vi
1,,.,,,.,,,,,.,.,,.,.,, [1].,,,. 1.1,.,.,,,,. 1
1.2 1.2,.,,,.,,,,.,,.,,. 1.3,.,,.,,.,,.,.,. 1.4,. 2,,,., 3,.,. 4,,,,. 2
2 2,.,,,.,.,,. 2.1,.,,,. 2.1.1,,,,,,,.,, Haar [2]., 2.1 2 ( -1, + 1).,.,. 3
2.2 2.1 2.1.2 Gallagjer [3].,., [4]., 2.1.1,.,. 2.1.3,.,,.,. 2.2,.,., MRF,. 4
2.2 2.2.1 MRF Markov Random Field, [5].,.,,., MRF,.,. 2.2, 2.3. 2.2 2.3 2.2.2,.,.,. 5
3 3,,.,,,,. 3.1,,.,.,,,,., 3.1. 3.2,,., [6]., I. I = Ī + αsin(2πωt) + kζ (3.1) Ī, α, ω, t, k(-1 k 1), ζ., I 6
3.3 3.1 Ī α ζ I Ī + α + ζ,., Ī α. Ī = n 1 n Ī + 1 n I (3.2) α = n 1 α + 1 2 n n (I Ī)2 (3.3) n,.,,, α. α = m 1 m Ī = Ī (3.4) α + 1 m 2 (I Ī)2 (3.5) m(m n),. 3.3.., [7]., Histograms of gradients(hog), ( ) ( ) [9]. 7
3.4 3.4. 3.2,,. 3.2 3.4.1,.,, Brent Berlin Paul Kay 1969.,,,,,,,,,,, 11. 11. 3.4.2, 3.3, RGB., RGB.,,,. 3.4, 3.5, 3.6. 8
3.4 3.3 3.4 R 9
3.4 3.5 G 3.6 B 10
3.5 3.5., 3.7,,,. 3.7 3.5.1.., Real Adaboost. Boosting,,. Real Adaboost, Adaboost. Adaboost, [8]. Real Adaboost, +1-1 2, W +, W., Real Adaboost,, 1. 11
3.5 3.5.2,.,., 3.8,,,. 3.8 3.5.3,, RGB. RGB,.,. 3.5.4,. RGB.,,. 12
3.5,,., 3.1, 9, 3., RGB 9, 3,. ( ) 2 3 4 8 76.2%,25.0% 73.3%,21.5% 55.6%,12.4% 9 78.5%, 28.4% 74.5%,17.0% 64.8%, 15.0% 10 81.2%,31.4% 78.2%,25.7% 66.3%,18.0% 3.1,, (, ) 13
4 3,. 4,,,,. 4.1,.,,, 2.,,.,,,. 4.1. 4.1 14
4.2 4.2,.,.,,,,,,. 4.1. 4.3 1,000.,, 74.8 %., 13.5 %., 61.7%, 18.9%., 74.5%, 17.0%. 74.8% 61.7% 74.5% 13.5% 18.9% 17.0% 4.1 4.4, 74.8%, 13.5%., 15
4.4,,.,.,,.,.,,. 16
5,,.,,., 75%., 14%.,,,,. 17
,,,,.. 18
[1],,,,, 2010. [2],,, 2007. [3] A.C.Gallagher et.al., Using Context to Recognie People in Consumer Images, IPSJ Transactions on Computer Vision and Applications, Vol.pp.115-126,2009. [4],,, 2009. [5], MRF,, 2009. [6],,,,,,, Vol.J88, No.5,pp.864-875,2005.5. [7],, Boosting,, 2008. [8], Adaboost,, 2006. [9], N.Dalal and B. Triggs. Histograms of Gradients for Human Detection, IEEE Computer Vision and Pattern Recognition, vol.1, pp.886-893, 2005. 19