Camera Motion Estimation and Optimization Approach

A camera motion estimation method based on feature matching for low-power processing platform is proposed in this paper. The proposed method is more effective than the previous methods that utilizes epipolar constraint, and provides a better time efficiency than epipolar constraint based method. Experimental results indicate the proposal saves 10% of the time consumption than previous method. In addition, this paper constructs a least squares model of camera re-projection error. On this basis, Levenburg-Marquart gradient descent method is employed to minimize the re-projection error. The experimental results show that the employed optimization method improves the accuracy of camera motion estimation about 10%.