中文摘要：

本文采用视觉显著性提出了一种移动机器人动态环境建模方法.该方法利用提出的视觉显著性模型,对连续的2帧图像中匹配的加速稳健特征点（SURF）利用其位置关系并采用多重随机抽样一致（multi-RANSAC）算法实现了环境中动态物体显著性检测.采用投影方法和快速均值漂移算法构建了动态环境的栅格模型,利用得到的动态显著性物体的位置更新环境地图中的栅格占据值以及动态物体的影响区域.动态环境显著图构建实验和动态环境的栅格模型构建实验的结果证明了上述方法是可行的.

英文摘要：

This paper presents a method of mobile robots dynamic environment modeling based on visual saliency.This method uses the proposed model of visual saliency to achieve the saliency detection of dynamic objects in the environment by using multi-random-sample-consensus（multi-RANSAC） algorithm and position relations of matching speededup-robust-feature（SURF） points in consecutive two images.It adopts the projection methods and fast mean shift algorithm to construct a dynamic environment grid model,the grid occupation area and the influence region of dynamic objects are updated by using position of the saliency objects.The results of dynamic environment saliency map construction experiments and dynamic environment grid model construction experiments illustrate the effectiveness of this method.