Generating a merged, fused three-dimensional point cloud based on captured images of a scene

1. A method for presenting a merged, fused three-dimensional point cloud, comprising:

acquiring a plurality of sets of images of a scene from different vantage points, each set of images including a respective plurality of stereo matched images and a color image;

for each respective set of images:

obtaining a disparity map based on the plurality of stereo images;

fusing data from the color image onto the disparity map so as to generate a fused disparity map; and

creating a three-dimensional fused point cloud from the fused disparity map; and

merging together each of the respective three-dimensional fused point clouds so as to obtain a merged, fused three-dimensional point cloud;

wherein fusing data from the color image onto the disparity map includes converting between color and grey-scale values.

2. The method of claim 1 wherein merging together each of the respective three-dimensional fused point clouds includes determining a correspondence between discrete ones of the three-dimensional fused point clouds based, at least in part, on contrast edge information obtained from the respective images.

3. The method of claim 1 wherein merging together each of the respective three-dimensional fused point clouds includes:

extracting respective two-dimensional key points from each of the three-dimensional color point clouds; and

identifying matching two-dimensional key points among different ones of the three-dimensional fused point clouds.

4. The method of claim 3 wherein the two-dimensional key points correspond to contrast edges.

5. The method of claim 1 wherein fusing data from the color image onto the disparity map further includes converting between high- and low resolution images.

6. The method of claim 1 including displaying the merged, fused three-dimensional point cloud on a display screen of a hand-held mobile device.

7. The method of claim 1 including displaying the merged, fused three-dimensional point cloud on a display screen of a computing device and, in response to receiving user input, rotating the merged, fused three-dimensional point cloud on the display screen.

8. The method of claim 1 including displaying the merged, fused three-dimensional point cloud on a display screen, wherein different depths or ranges of depth are displayed as different colors or using other visual indicators to differentiate between the different depths or ranges of depth.

9. The method of claim 1 further including performing an interpolation process to reduce sparseness of the disparity map or the fused disparity map, the interpolation process including:

scanning the disparity map or the fused disparity map using a block of specified size; and

updating a disparity value for a designated pixel of each particular block during the scanning if the designated pixel of the particular block has a current disparity value that is invalid.

10. The method of claim 9 wherein updating a disparity value for a designated pixel of each particular block preserves geometric edges of the disparity map or the fused disparity map.

11. The method of claim 9 wherein updating a disparity value for a designated pixel of each particular block if the designated pixel has a current disparity value that is invalid includes:

if the particular block has a number of valid pixels such that a difference between maximum and minimum disparity values in the particular block is smaller than a second predefined threshold, assigning a mean disparity value to the designated pixel in the particular block; and

if the particular block has a number of valid pixels such that a difference between the maximum and minimum disparity values in the particular block is larger than the second predefined threshold, assigning to the designated pixel of the particular block a disparity value equal to a most frequent disparity value from among a predefined subset of disparity values in the particular block.

12. A method of converting a relatively sparse disparity map into a less sparse disparity map, the method comprising:

scanning the disparity map using a block of specified size; and

updating a disparity value for a designated pixel of each particular block during the scanning in response to the designated pixel of the particular block having a current disparity value that is invalid,

wherein updating the disparity value includes

assigning a mean disparity value to the designated pixel in the particular block when the particular block has a number of valid pixels such that a difference between maximum and minimum disparity values in the particular block is smaller than a second predefined threshold, and

assigning to the designated pixel of the particular block a disparity value equal to a most frequent disparity value from among a predefined subset of disparity values in the particular block when the particular block has a number of valid pixels such that a difference between the maximum and minimum disparity values in the particular block is larger than the second predefined threshold;

wherein updating the disparity value for the designated pixel of each particular block preserves geometric edges of the disparity map.

13. The method of claim 1 wherein the sets of images are acquired using a plurality of depth channels including infra-red sensors to acquire stereo matched images and a third channel including a RGB sensor to acquire color images.

14. The method of claim 1 further including performing an interpolation process to reduce sparseness of the disparity map or the fused disparity map, the interpolation process including:

scanning the disparity map or the fused disparity map using a specified block size;

updating a disparity value of a designated pixel in a particular block, wherein the updating includes:

determining whether a number of pixels in the particular block having a valid disparity value is less than a first threshold value, and

if the number of valid disparity values in the particular block is greater than a first threshold value, determining whether a standard deviation for disparity values of pixels in the particular block is less than a second threshold value, and

if the standard deviation for disparity values of pixels in the particular block is less than the second threshold value, assigning a most frequent disparity value in the particular block to the designated pixel.

15. The method of claim 14 including repeating said updating for each of a plurality of blocks of pixels.

16. The method of claim 1 further including performing an interpolation process to reduce sparseness of the disparity map or the fused disparity map, the interpolation process including:

scanning the disparity map or the fused disparity map using a specified block size;

updating a disparity value of a designated pixel in a particular block, wherein the updating includes:

determining whether a number of pixels in the particular block having a valid disparity value is less than a first threshold value, and

if the number of pixels in the particular block having a valid disparity value is less than a first threshold value, assigning a value of zero to the designated pixel, and

if the number of valid disparity values in the particular block is greater than a first threshold value, determining whether a standard deviation for disparity values of pixels in the particular block is less than a second threshold value,

wherein if the standard deviation for disparity values of pixels in the particular block is less than the second threshold value, assigning a most frequent disparity value in the particular block to the designated pixel, and

if the standard deviation for disparity values of pixels in the particular block is not less than the second threshold value, assigning a mean disparity value for the pixels in the particular block to the designated pixel.

17. The method of claim 16 including repeating said updating for each of a plurality of blocks of pixels.

18. The method of claim 9 wherein in response to the designated pixel having a current disparity value that is invalid and a number of valid pixels in the particular block is smaller than a first predefined value, then leaving the disparity value of the designated pixel of the particular block as invalid.