• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Intelligent robots and computer vision XXXI: algorithms and techniques >Real time moving object detection using motor signal and depth map for robot car
【24h】

Real time moving object detection using motor signal and depth map for robot car

机译：利用电机信号和深度图对机器人小车进行实时运动物体检测

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Moving object detection from a moving camera is a fundamental task in many applications. For the moving robot car vision, the background movement is 3D motion structure in nature. In this situation, the conventional moving object detection algorithm cannot be use to handle the 3D background modeling effectively and efficiently. In this paper, a novel scheme is proposed by utilizing the motor control signal and depth map obtained from a stereo camera to model the perspective transform matrix between different frames under a moving camera. In our approach, the coordinate relationship between frames during camera moving is modeled by a perspective transform matrix which is obtained by using current motor control signals and the pixel depth value. Hence, the relationship between a static background pixel and the moving foreground corresponding to the camera motion can be related by a perspective matrix. To enhance the robustness of classification, we allowed a tolerance range during the perspective transform matrix prediction and used multi-reference frames to classify the pixel on current frame. The proposed scheme has been found to be able to detect moving objects for our moving robot car efficiently. Different from conventional approaches, our method can model the moving background in 3D structure, without online model training. More importantly, the computational complexity and memory requirement are low making it possible to implement this scheme in real-time, which is even valuable for a robot vision system.
机译：在许多应用中，从移动摄像机检测运动对象是一项基本任务。对于移动的机器人汽车视觉，背景运动本质上是3D运动结构。在这种情况下，常规的运动对象检测算法不能用于有效地处理3D背景建模。在本文中，提出了一种新颖的方案，该方案利用从立体摄像机获得的电机控制信号和深度图对运动摄像机下不同帧之间的透视变换矩阵进行建模。在我们的方法中，通过使用当前电机控制信号和像素深度值获得的透视变换矩阵对摄像机移动过程中帧之间的坐标关系进行建模。因此，可以通过透视矩阵将静态背景像素与对应于相机运动的运动前景之间的关系相关联。为了增强分类的鲁棒性，我们在透视变换矩阵预测期间允许了公差范围，并使用多参考帧对当前帧上的像素进行分类。已经发现提出的方案能够有效地检测我们的移动机器人汽车的移动物体。与传统方法不同，我们的方法无需在线模型训练即可在3D结构中对运动背景进行建模。更重要的是，计算复杂度和内存需求低，使得可以实时实施此方案，这对于机器人视觉系统而言甚至很有价值。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号