Human Action Recognition using Image Processing and Artificial Neural Networks

Human action recognition is an important technique and has drawn the attention of many researchers due to its varying applications such as security systems, medical systems, entertainment. Action recognition is an interesting and a challenging topic of computer vision research due to its prospective use in proactive computing. The developed algorithm for the human action recognition system, which uses the two-dimensional discrete cosine transform(2D-DCT) for image compression and the self organizing map(SOM) neural network for recognition purpose, is simulated in MATLAB. By using 2D-DCT we extract image vectors and these vectors become the input to neural network classifier, which uses self organizing map algorithm to recognize elementary actions from the images (trained). In this paper we have developed and illustrated a recognition system for human actions using a novel self organizing map based retrieval system. SOM has good feature extracting property due to its topological ordering. Using an image database of 30 action images, containing six subjects and each subject having five images with different body postures reflects that the action recognition rate using one of the neural network algorithm SOM is 98.16%. General Terms Human Action Recognition (HAR), Artificial Neural Network (ANN).