Action Recognition by Converting Normal Video into Silhouette Video

Nowadays action recognition has many applications such as robotics, human computer interface etc. This project proposes a method for action recognition that uses a number of correlated poses. Here normal video is converting to silhouette video and then each silhouette video is converting to a set of images frames for feature extraction. The silhouettes are used as input features for the BoCP ( Bag of Correlated Poses) model. BoCP are using for encoding the local features of action. After the descriptor extraction C-means clustering algorithm is using to generating the codebook. Clustering is the process of dividing the dataset into subsets, so that the data in each subset shares some common values according to some defined distance measure. To reduce the high dimensionality of computed feature Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) are using. Experimental results prove the viability of the complementary properties of two descriptors and the proposed approach outperforms the state of the art methods on the IXMAS action recognition dataset.