Electronic digital image stabilization is a video image sequence stabilization technology which has become a key point in manual arms and equipment, especially in the military field. Due to impact of carrier stance and vibration of the airborne imaging system, the image sequence changes hard between frames that causes observers feel vague and unstable and is one of the bottlenecks in air investigation, target detection, assessment of attack effects.In this paper, a digital image stabilization system based on Gray-coded bit-plane algorithm has been studied. This rapid motion vector estimation process through XOR bit operation greatly reduces the computational complexity, while a mean filter method is used to determine the motion compensation of each frame that could smooth the image sequence output. This stabilization system is a real-time system which is easy to hardware implementation.Compared with the translation model, the affine model with 6 parameters has a greater accuracy in describing geometric deformation between frames. This paper presents a digital image stabilization algorithm based on affine model that could get motion parameters by affine transform in the process of motion vector estimation. The new stabilization algorithm is robuster and more accurate in detecting the rotating, zooming, and deformation between frames than block matching algorithm.Experiments of image stabilization are taken on some real image sequences in the end of the paper, applying the stabilization algorithms which are based on Gray-coded bit-plane and affine model. The stabilization effects of the two algorithms are evaluated, and the performances and the application scopes are analyzed. |