3D, point cloud splicing

As one of the hot research directions of point cloud , Point cloud splicing has always been the object of many scientific researchers . that , What is the charm of it ？

Before the official introduction , We might as well ask ourselves a few questions ：

• Point cloud splicing 、 What is the relationship between registration and registration ？
• Why do you want to do point cloud splicing ？
• How to do some cloud splicing , How to judge the quality of splicing ？

If you can answer the above three questions accurately , Please go home , Because watching it again may waste your time . If you can't answer , Please take your seat ~

1、 Point cloud splicing , Registration , What is the connection between registration ？

These are three words we often see , Some papers or books are called splicing 、 Some are called registration , Some foreign software may use the word "registration" , I can't tell whether it's stupid or not . If I say these three are the same concept , Will you hit me , In fact, the three can be replaced by the same English ,Point Cloud Registration, It's just that the translation is different ~, namely The point cloud of different positions is passed through the information of the overlapping part , Change to the same location . Because I personally prefer to use splicing （ A good understanding ） The word , therefore , Next, we will use the term "splicing" to describe , Looking forward to the sea culvert .

2、 Why do you want to splice point clouds ？

Here's a small example ：

I'm in a room , Hold RGB-D The camera took a group of photos . Unlike ordinary cameras , Each pixel in these photos has both color and depth . The effect of this depth value is not trivial , We can combine these photos according to the shooting angle and shooting position of each photo , Form a three-dimensional pattern . The five pictures shown in the following figure are examples , They were shot from different angles