Perspective distorted video restoration and stabilization for mobile devices

Pictures/Videos taken by mobile devices may suffer degradation problems such as perspective distortion, noises and shakes. In this paper, we present a novel approach to solve the perspective distortion and camera shake issues. A perspective correction approach is proposed based on a pyramidal quadrangle detection method. An adaptive window-based autoregressive(AR) interpolation is brought up to enhance the warping result. We rectify the transformation model with smoothing motion vector to realize a video stabilization process that improves the video quality. Experimental results confirm the enhancement of the visual quality our method brings.