ICP 算法简介

• 由点云配准 说起

  　Point cloud registration, is the process of finding a spatial transformation that aligns two point clouds. The purpose is to merge point clouds of multiple views into a globally consistent model.
  　Iterative Closest Points (ICP) is an algorithm employed to minimize the difference between two clouds of points. In the algorithm, target point cloud, is kept fixed, while the other one, the source, is transformed to best match the reference (the target). The algorithm iteratively revises the transformation (combination of translation and rotation) needed to minimize the distance from the source to the reference point cloud.

　情况是这样的，在之前讲过的回播 .pcap 数据的 Velodyne_player 程序中，需要调用 Winpcap (其实就是 libpcap 的 Win挫版) 的 API 解析 .pcap 数据，再通过 UDP 发送出去。我们的 Velodyne_player 是一个 Win32 的程序，显然调用的就是32位的 Winpcap 库的 API； 后来我们也移植了一个 .pcap 采集程序的 Linux 版本，结果，用该 Linux 版本采集程序采集到的 .pcap 数据却没办法用我们 Win 下的 Velodyne_player 回播。后来发现，我们的 Linux 版本的采集程序用的是64位的 libpcap 库(因为系统是64位的 Ubuntu16.04，默认安装的就是64位的 libpcap 库)，64位和32位的 libpcap，在时间戳上有很关键的区别，下面是开源的 pcap.h 中的声明:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

/*
 * Generic per-packet information, as supplied by libpcap.
 *
 * The time stamp can and should be a "struct timeval", regardless of
 * whether your system supports 32-bit tv_sec in "struct timeval",
 * 64-bit tv_sec in "struct timeval", or both if it supports both 32-bit
 * and 64-bit applications.  The on-disk format of savefiles uses 32-bit
 * tv_sec (and tv_usec); this structure is irrelevant to that.  32-bit
 * and 64-bit versions of libpcap, even if they're on the same platform,
 * should supply the appropriate version of "struct timeval", even if
 * that's not what the underlying packet capture mechanism supplies.
 */
struct pcap_pkthdr {
  struct timeval ts;	/* time stamp */
  bpf_u_int32 caplen;	/* length of portion present */
  bpf_u_int32 len;	/* length this packet (off wire) */
};