基于协同显著性的服务机器人空间物体快速定位方法

发布时间：2018-01-14 01:21

  本文关键词：基于协同显著性的服务机器人空间物体快速定位方法　出处：《机器人》2017年03期 　论文类型：期刊论文

  更多相关文章： 服务机器人 定位方法 协同显著性 图像间显著性传播 流形排序

【摘要】：针对传统算法模型先检测、后识别、再定位导致执行效率较差的问题,提出了一种基于协同显著性检测的服务机器人空间物体快速定位方法.利用RGB-D传感器获取N对包含待定位物体的RGB图像与深度图像,将待定位物体看作协同显著性目标,在RGB图像中充分挖掘单幅图像显著性传播机理,构建基于图像间显著性传播和图像内流形排序的两阶段引导协同显著性检测模型,同时排除背景和非协同显著性物体,得到协同显著性物体区域的像素坐标集合.进一步利用RGB图像与深度图像的对应关系确定物体质心的空间坐标,实现对空间物体的快速定位.最后将所提方法在iCoseg标准数据库和经手眼标定后的服务机器人机械臂抓取平台上进行实验.实验结果表明,该方法一致优于现有的5种协同显著性检测算法,且满足抓取系统的实时性需求,在复杂背景、多目标干扰以及光照变化时具有较高的准确性和鲁棒性.
[Abstract]:The traditional algorithm model is detected first, then identified, and then located, which leads to poor execution efficiency. This paper presents a fast localization method for space objects of service robots based on cooperative salience detection. The RGB-D sensor is used to obtain N pairs of RGB images and depth images containing the objects to be located. The object to be located is regarded as a cooperative significant target, and the significant propagation mechanism of a single image is fully mined in RGB images. A two-stage guided cooperative saliency detection model based on image salience propagation and image manifold sorting was constructed, while background and non-cooperative salience objects were excluded. The pixel coordinate set of the synergetic significant object region is obtained, and the spatial coordinate of the center of mass of the object is determined by using the corresponding relationship between the RGB image and the depth image. Finally, the proposed method is tested on the iCoseg standard database and the manipulator grab platform after the hand-eye calibration. The experimental results show that the proposed method can be used to locate the space objects quickly. The proposed method is superior to the five existing cooperative saliency detection algorithms, and meets the real-time requirements of the grab system. It has high accuracy and robustness in complex background, multi-target interference and illumination change.
【作者单位】： 北京工业大学信息学部;河南科技学院机电学院;
【基金】：北京工业大学2017智能制造领域大科研推进计划(040000546317552) 国家自然科学基金(61175087)
【分类号】：TP242
【正文快照】： 1引言(Introduction) 制约智能服务机器人发展的关键技术问题就是如何使机器人具有人类的智慧,其中对环境物体准确、快速的空间感知又是亟待解决的首要问题.近年来,针对不同视觉传感器的物体检测、识别、定位技术一直都是机器视觉领域的研究热点[1-3]. 服务机器人通过自身搭，

本文编号：1421411