基于OpenCV的摄像头动态手势轨迹识别及其应用

基于OpenCV的摄像头动态手势轨迹识别及其应用，前沿的研究，最新论文2012B_0% random t XoX,XX4_1、42、B0Bhattacharyyaglsl. 8 gsl_ran_gaussianpp q1/3t=∑∵n8HSVcerterHSVIISV V44 HMM3IIMMA=SO4 B450AiI i j-+∑bk2HMM4+1=A_1*X1-X0+A_2*Xo1994-2013ChinaAcademicJournalElectronicPublishingHouse.Allrightsreservedhttp://www.cnki.netOpencvlIMMP., v3)A BP(IAIIMM(a)两点之间的方向夹角(b)轨迹的方间夹角分为8等份HMMBaum-WelchyO aB,i/∑a,iBtHMMAHMM∑s: t Hq,=======-s,qui a 6. o, B, 1 j/PolAa)向右的于势(b)向左的手势(c)向上的手势(d向下的手势动,O:B1j1+1J∑,ijtimermero∑点:1j/∑b kFirmerlimmer=3∑y,1/∑yIMM丌AB512Baum-welch4.3HMⅥHMMHMⅥHMM8(a)第5(b)第6帧第6幀第6帧o1994-2013ChinaAcademicJournalElectronicPublishingHouse.Allrightsreservedhttp://www.cnki.net1322012(a第1(b)第72(a)第5帧(b)第60帧(c}第62帧(d)第G3帧(c)第75d)第呲5.4 HMMHMMHMMHMIHMM丌ABHMMa)第4帧.0()第50d)第54帧HMMHMM0.2928060.56l9010.2943160.5606530.1167910.1900640.0428680.1903360.041360.3217210.3235800.0173120.1166050.0155030.0708480.069680.1601290.0412690.1600400.039740.0749660.0738510.1859570.1468650.1466160.2136800.1587320.0954920.1586260.094620.2069880.207462HMM30.00l20.00100.0010.00I0.00l0.000.0050.00l.O001.000360.0010.00120.00I0.0010.00l0.001IIMMHMMIIMM320×240HMMo1994-2013ChinaAcademicJournalElectronicPublishingHouse.Allrightsreservedhttp://www.cnki.netOpencv133jck hand recogni tion2005,2611):29Openly获取直方图轨迹图像2010,30(71:59-61.在单击“轨迹「未选择-图像显示[5 QIN WEN, PENG QICONG. An improved particle filter algorithm图像时选择Chmm控制based on neural network for visual tracking[c / International Con个按钮选C控制贪食鱼ference on Communications, Circuits and SyLas v.项训练HMMIEEE.2007:765-768LD」停止退出2009[7 YU JINXIA, LIU WENJING, YANG Y. Impmved particle filter algorithms based on partial systematic resampling C|//IEEE InTernational Conference on Intelligent Computing and Intelligent SystemsLas vegas: IEEE, 2010: 483-4878 FAZLI S, POUR II M, BOUZARI ll. Particle filter based objecttracking with sift and color feature[C // Second International Conas Vegas: IEEE, 2009: 89-9(8:62-64.L10」LD」HMM,2004[ II YUAN LICHI. An improved HMM speech recognition model[c//International Conference on AudiO. las Vegas: IEEE, 2008: 1311LlOpenCVLD」1315.[12 TAN WENJUN, WU CHENGDONG, ZHAO SHUYING. Recogni[2 CHlO JU, JIN S ll, PILAM X D. Object tracking circuit using partition of dynamic hand gesture based on SCllMM model[C]//Chi-cle filter with multiple features [C//SICE-ICASE: Internationalnese Control and Decision Conference. Log Vegas: IEEE. 2009Joint Conference. Las Vegas: IEEE, 2006: 1431-14362430-24341272011,28(2):242-244.C,2010,1(1):92-95SIFT[8 BEIS J S, LOWE D G. Shape indexing using approximate nearest-neighbour search in high-dimensional spaces [C]// Proceedings ofthe IEEE 1997 Computer Society Conference on Computer visionand Pattern Recognition. Washington, DC: IEEE Computer Socie-ly,997:1000-1006SIFT009,29(5):423-2262009,30(4):857-861[O][2] LOWE D G. Distinctive image features from scale-invariant key-points[ J]. International of Computer Vision, 2004, 60(2):91[11 SHUM H Y, SZELISKI R. Construction of panoramic image mosaicswith global and local alignment[ J International Journal of Comp3 MIKOLAJCZYK K, SClIMID C. A performance evaluation of localer visian,2000,36(2):101-130descriptors[ J|. IEEE Transactions on Pattern Analysis and Machine[12 ZOGIILAMI 1. FAUGERLAS O, DERICIIE R. Using geometric corIntelligence,2005,27(10):1615-1630ners to build a 2D mosaic from a set of images[C]// Proceedings of4 LINDEBERG T. Seale -space theory a basic foul for analyzingthe International Conference on Computer Vision and Pattern Recog-structures at different seales LJ. Journal of Applied StatisticitiL. Washington, DC: IEEE Computer Society, 1997: 1671994,21(2):224-270o1994-2013ChinaAcademicJournalElectronicPublishingHouse.Allrightsreservedhttp://www.cnki.net