发明名称 SYSTEMS AND METHODS FOR 3D POINT CLOUD PROCESSING
摘要 Generally discussed herein are systems, apparatuses, and methods for moving object detection. In one or more embodiments, a method can include computing a central value for each color of each pixel in a series of frames from a static imaging device to create a static reference frame, computing a difference between pixel values of a first frame captured after the series of frames by the static imaging device and a central value of a corresponding pixel value in the static reference frame, comparing the computed differences to a difference threshold, aggregating contiguous pixels corresponding to differences greater than the difference threshold into contiguous segments of pixels, and determining if the contiguous segments correspond to a moving object by registering the contiguous segments to pixels of a frame preceding the first frame.
申请公布号 US2016321820(A1) 申请公布日期 2016.11.03
申请号 US201514701801 申请日期 2015.05.01
申请人 Raytheon Company 发明人 Ely Richard W.
分类号 G06T7/20;G06T7/00 主分类号 G06T7/20
代理机构 代理人
主权项 1. A system for three dimensional point cloud processing comprising: processing circuitry configured to execute instructions to implement modules comprising: a static scene creation module to compute a plurality of central values to create a static reference frame, one central value for each of a plurality of colors for each pixel in a series of frames from a static imaging device, each frame in the series of frames including pixel values of a scene captured by the static imaging device;an object detection module to compute a difference between each of a plurality of pixel values of a first frame captured by the static imaging device after the series of frames and a corresponding central value of a corresponding pixel in the static reference frame, compare the differences to a threshold, and aggregate contiguous pixels corresponding to differences greater than the threshold into contiguous segments of pixels;a confirmation module to determine if the contiguous segments correspond to a moving object by registering pixels of the frame to pixels of a frame preceding the first frame; and a three dimensional point cloud extraction module to update a three dimensional point cloud using the contiguous segments determined to correspond to the moving object.
地址 Waltham MA US