Vector quantization in HMM

2 次查看（过去 30 天）
Sophie
Sophie 2016-3-23
hello, I work with Hidden Markov models for gesture recognition, i had a problem in the data quantization step of each feature vector . I had the data training = (f1, f2, f3; .....) fi correponds to the feature vector at time i. we suppose that the feature vector is the center of the hand (x, y, z) and the speed of the hand movement (vx, vy, vz) so the feature vector is (x, y, z, vx, vy, vz) I would like to quantify this vector into a symbol o0, I found in the net that i should use the kmeans algorithm, if it's true, the kmeans function has as inputs ( k the number of cluster and obs a set of observations vectors to cluster ) so in this case obs decomes the following matrix: obs = [x, y, z, vx, vy, vz; x1, y1, z1, VX1, VY1, VZ1; x2, y2, z2, vx2, vy2, VZ2; ...]? with [x, y, z, vx, vy, vz] the feature vector at time t1 (frame1) [x1, y1, z1, VX1, VY1, VZ1] the feature vector at time t2 (frame2)... the output will be the vector K[K1,K2,K3...] k1 corresponds to symbol o0, k1 to symbol o1... please I would know if what I said just now is correct or not.
Thanks in advance

请先登录，再回答此问题。

回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Statistics and Machine Learning Toolbox 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by