Feature-based Video Stabilization using Gabor Wavelets

This study proposes a method to stabilize jittery video using a feature-based technique. our feature-based technique extracts local image features using gabor wavelets. firstly,to locate a set of interest points within a video frame,we detect some local maxima on gabor response map image. then,using the same gabor response map image,we compute relational features around these interest points. the method was tested using shaky car video obtained from matlab version 2011b and compared with the sift and surf methods. the output of using the proposed local image features is comparable to the output produced by sift and surf methods and has shown good result concerning stabilization and discarded distortion from the output video.