Object tracking and matching in a video stream based on SURF and wavelet transform

In computer vision, visual object tracking is a significant task for monitoring applications. Tracking of object type is a matching trouble. In object tracking, one main difficulty is to select features and build models which are convenient for distinguishing and tracing the target. The suggested system for continuous features descriptor and matching in video has three steps. Firstly, apply wavelet transform on image using Haar filter. Secondly interest points were detected from wavelet image using features from accelerated segment test (FAST) corner detection. Thirdly those points were descripted using Speeded Up Robust Features (SURF). The algorithm of Speeded Up Robust Features (SURF) has been employed and implemented for object in video stream tracking and matching. The descriptor of feature in SURF can be operated by minimizing the space of search for potential points of interest inside the scale space image pyramid. The tracked interest points that are resulted are more recurrence and pother free. For dealing with images that contain blurring and rotation, SURF is best. Fast corner detector can be employed along SURF method to build integral images .The integral images can be used to enhance the speed of image matching. The features that are extracted from video images are matched using Manhattan distance measure. Apply the algorithm of FAST corner detection along SURF descriptor of feature; tracking and matching adequacy is better, fast and more efficient than Scale Invariant Feature Transform SIFT descriptor. The experimental outcomes displayed that the time that SURF could be taken for matching is less than the time that SIFT could be taken ,the SURF accuracy depends on number of key-points which are extracted from each frame. SURF key-points are less than SIFT key-points; therefore, SURF key-points could be considered optimal in the process of matching accuracy.