Even though the development of computer's input device has made life easier in different aspects, many people still finds interaction with computers a difficult task. This human desire ushered in the communication with computers in human's natural language like speech and body language. Our project is an example of it. The aim of this project is to implement hand gesture recognition, using general purpose hardware and inexpensive sensors like simple personal computer, USB web camera, in real time vision based system for the implementation of visual interaction with computers in home or office.
The overall project can be categorized into four different tasks i.e. acquisition of image from a web camera, quick segmentation of the hand from image, recognition of the hand gesture, and implementation of the command depending on the hand gesture. The fundamental concept of our project is the segmentation of the moving hand from the real time video irrespective of the background of image and the lightning condition. Many hand postures are to be included in this project that would make the computer human interface more easy and accurate. The characteristics shape of the hand is utilized to detect the hand in an image during segmentation. The contour of the hand represents the shape of the hand and is therefore independent of the viewpoint, skin color and illumination of the image.
In our project the segmentation is carried out by background subtraction of two image frame in RGB scale; the subtracted image is converted to the Grayscale and then the image is binarized using a threshold value. The contour is used to extract the rectangular section from the binary image, the rectangular image is re-scaled to 30 * 30 pixels image. These images are stored from the real time video stream as the data set. The classification and recognition of the features are done using Neural Network. The concept of the project can be applied to various applications like Sign language application, controlling computer operated machines. Our project is build around PYTHON and MATLAB.
Technology and Programming paradigm used: Python, Matlab, Image Processing, Realtime