rgbd::registerDepth segfault when depth resolution > rgb resolution

[NickLaurenson-Visionick]

System information (version)

• OpenCV => 4.10.0
• Operating System / Platform => ubuntu 22.04
• Compiler => g++-11

Detailed description

cv::rgbd::registerDepth segfault in case the depth resolution is greater than the rgb resolution.

This is caused by a wrong argument used in the contructor here, and could be fixed as follow:

--- a/modules/rgbd/src/depth_registration.cpp
+++ b/modules/rgbd/src/depth_registration.cpp
@@ -146,7 +146,7 @@ namespace rgbd
         // Apply the initial projection to the input depth
         Mat_<Point3f> transformedCloud;
         {
-            Mat_<Point3f> point_tmp(outputImagePlaneSize,Point3f(0.,0.,0.));
+            Mat_<Point3f> point_tmp(unregisteredDepth.size(),Point3f(0.,0.,0.));
 
             for(int j = 0; j < unregisteredDepth.rows; ++j)
             {

Steps to reproduce

#include <opencv2/opencv.hpp>
#include <opencv2/rgbd.hpp>

int main() {
  // Example of a 1000x1000 depth image we want to register into a 600x600 color image

  // Define the intrinsic matrices
  cv::Matx33d intrinsics_depth(100, 0, 500, 0, 100, 500, 0, 0, 1);
  cv::Matx33d intrinsics_color(100, 0, 300, 0, 100, 300, 0, 0, 1);
  cv::Mat dist_coef_color; // Empty distortion coefficients
  cv::Matx44d extrinsics = cv::Matx44d::eye();

  // Create a depth image with resolution 1000x1000
  cv::Mat depth(1000, 1000, CV_64F);

  // Define the output resolution
  int width = 300;
  int height = 300;

  // Register depth
  cv::Mat registered_depth(cv::Size(width, height), CV_64F);

  cv::rgbd::registerDepth(intrinsics_depth,
                          intrinsics_color,
                          dist_coef_color,
                          extrinsics,
                          depth,
                          cv::Size(width, height),
                          registered_depth,
                          false // Depth dilation
  );

  return 0;
}

Issue submission checklist

• I report the issue, it's not a question
• I checked the problem with documentation, FAQ, open issues,
  forum.opencv.org, Stack Overflow, etc and have not found any solution
• I updated to the latest OpenCV version and the issue is still there
• There is reproducer code and related data files: videos, images, onnx, etc