Research on the Application of Visual SLAM in Embedded GPU

Algorithm 3

Local point selection algorithm.

Input: Feature point information in point cloud maps inInf[0…pointNum]

Output: Selected feature point information outInf[0…pointNum], Number of feature points selected nSelection

1 cudaMemCopyHostToDevice(d_inInf[0…pointNum], inInf[0…pointNum]);

2 for i 0 to pointNum

3  if(gpuIsSelection(d_inInf[i]))

4   d_outInf[i].flag=0;

5   d_outInf[i] gpuSaveSelectedPointsInformation();

6  end

7 end

8 cudaMemCopyDeviceToHost(outInf[0…pointNum], d_outInf[0…pointNum]);

9 nSelection=0;

10 for i 0 to pointNum

11  if(outInf[i].flag==0)

12   nSelection++;

13  end

14 end

Note: The beginning of d refers to the data allocated in the video memory, pointNum is the number of feature points in the point cloud map