MULTIPLE OBJECT TRACKING EMPLOYING OPTIMAL PARAMETRIC ESTIMATION

Abstract

The area of Object Tracking is of immense interest due to its application in automated surveillance, motion based recognition, pedestrian monitoring, human computer interaction. Visual tracking is the process of repeated estimation of the state of an object in the next frame, given states in previous frames. Object identification is one of the initial, but paramount steps in object tracking. It is basically, determination of video statistics, object classification, determination of inconsistencies, and then finally human identification. Every object has its unique features in a video scene. These unique features help us to determine whether the object is same in the next frame of a video as we need to track or not. In our project we have first extracted those features using communication theory, or radar theory to be precise. This enriches us with the crucial estimated features of the objects. After this parametric estimation, we have calculated the optimum detection probabilities for the target object using Neyman Pearson Theory. Considering the decision probabilities, the miss probability, the false probability, the detection probability and the minimized cost function, determined using Neyman Pearson Theory, we identify the target object.