생물학 기반의 확률적 모션 추정 모델을 이용한 영상 분할

Abstract

In this Paper we present a model for image segmentation taking into consideration motion and color. We divide the image into segments in first stage to resolve the problem of two or more object having similar color values being in one segment. Additionally a textured object divided into segments initially is merged with one segment by using motion information. For reliability and accuracy, we developed a bio-inspired motion estimation model based on the Bayesian formulation. In this model, motion is estimated by probabilistic processing of the pixel and region layer. Several experiments were performed to evaluate the performance of our model. The results of these experiments showed that the proposed methodology is effective for image segmentation.