🟧 Point Cloud Generate 2D ROI

Input an ordered point cloud, the operator analyzes the depth values of its Z-channel, automatically finding the minimum rectangular region that all valid points (Z values not 0 and not NaN) occupy on the 2D image plane, which is the generated 2D ROI.

When a target object point cloud is segmented from the scene, and further processing of that object in the 2D image is needed, this operator can map the 3D spatial target position back to the 2D image region.