You might try automated data registration in cloud compare which is based on ICP (Iterative closest point) to bring the point cloud in the same coordinate system. In this case, the results depends on the initial orientations of the point cloud. If initial orientations are close to each other, then you might achieve a better result. You can do it manually by selecting a minimum of four well-distributed and recognisable points in cloud compare or you can go for more points for a least-squares solution. You may also want to try open-source code based on normal distributions transform (NDT) for registering the two data together.

Sarvesh Kumar Singh thank you for ur answer! Could u go more specific about NDT?

Hi Nikita,

NDT does not require initial alignment of point cloud and still gives better data registration although processing time is more in comparison to ICP. I would suggest using point cloud library for data stitching. It is open source and you can use the code straightaway. Here is the link to NDT and its implementation in point cloud library. If you are not sure how to use it then probably start with cloud compare and use ICP after roughly aligning your point cloud.

http://www.pointclouds.org/documentation/tutorials/normal_distributions_transform.php

https://ieeexplore.ieee.org/document/7260213

Here is MATLAB implementation of NDT

https://au.mathworks.com/help/vision/ref/pcregisterndt.html

MATLAB implementation of ICP

https://au.mathworks.com/help/vision/ref/pcregrigid.html

To achieve more accurate result you can use local point descriptors provided in point cloud library for data registration or point cloud stitching.

Regards,

Sarvesh