A Fusion Method of Point-Cloud Based on Parallel Estimation of Distribution Algorithm

Abstract

The paper proposes a new method that utilizes Parallel Estimation of Distribution (PED) to obtain disparity map of two images and fuses 3D point-cloud by the disparity. Estimation of Distribution (ED) has several advantages such as low complexity and high efficiency while it also has the shortcoming that it’s sensitive to initial samples and the final solution is local optimum but global optimum. In order to exert its merit and overcome the shortcoming, this paper will improve the ED by parallel sampling to diminish the sensitivity with Adaptive Support-Weight method. We called it Parallel Estimation of Distribution (PED). After disparity map is obtained, the two images will be divided into high frequency and low frequency by lifting wavelet respectively and the frequency coefficient of each image will be averaged into fusion frequency. Finally, the fused frequency coefficient will be transformed by inverse lifting wavelet to the final fused image. The experiment has proved that the matching speed is outstanding without losing precision.