Tracking of a target person using face recognition by surveillance robot

In this paper we designed an experimental setup in order to have human-robot interaction i.e. first we are going to detect the face and after that we recognise the detected face. Afterwards we get the persons upper body torso color as a key feature. As we extracted the color feature we can compute the moments and also evaluate the motion parameters so that the surveillance robot can track the person accordingly. We also had introduced Speech module in order to have a interaction between the remote and base station. Surveillance robot must track the targeted person in a robust manner in indoor and outdoor environment in different light and dynamic varying conditions. In our proposed setup we use PCA which is going to recognise the person in a real time environment and should communicate to the person via speech module deployed in the surveillance robot, as face recognition works on real time environment we are getting average recognition rate of 98%. Experiment demonstration validates the efficient performance of the approach.