Realogram scene analysis of images: multiples for scene analysis
The techniques include an image recognition system to receive a realogram image including a plurality of organized objects and to detect and identify objects in the realogram image of one or more items on a retail shelf, identify shelf fronts and labels on the shelf fronts, identify empty space under shelves, identify areas where unidentified products may be, and identify areas where products are “out of stock”.
1. A computer-implemented method for identifying multiple identical objects in an image of a plurality of organized objects, the method comprising:
receiving an image of the plurality of organized objects;
extract feature points from the image of the plurality of organized objects;
finding matching feature points in the image of the plurality of organized objects;
clustering matching feature points with a matching distance and angle between feature points;
finding multiple identical objects based on the clusters by:
finding a statistical measure of vertical distances between matching feature points in the clusters;
finding a maximum number of identical objects in a stack based on the statistical measure of vertical distances;
determining coordinates for each of the identical objects in the stack with a height corresponding to the statistical measure of vertical distances; and
determining a number of stacks of the identical objects based on matching feature points that are horizontally aligned.
2. The computer-implemented method of claim 1 , further comprising locating unindexed objects using groups of identical objects that don't correspond to identified indexed objects.
3. The computer-implemented method of claim 1 , wherein finding matching feature points in the image comprises using a tree approach.
4. The computer-implemented method of claim 1 , further comprising:
identifying feature points in a cluster that are vertically aligned;
identifying feature points in the cluster that are horizontally aligned; and
identifying an area surrounding the vertically and horizontally aligned feature points as an area of unindexed products.
5. The computer-implemented method of claim 1 , further comprising:
generating image segments based on groups of pixels in the image of the plurality of organized objects;
labeling the image segments with cluster information; and
wherein finding multiple identical objects based on the clusters includes identifying image segments labeled with a cluster that have a consistent spatial relationship with other image segments labeled with the cluster.
6. The computer-implemented method of claim 1 , wherein finding matching feature points comprises identifying feature points that have a descriptor difference within a threshold.
7. The computer-implemented method of claim 1 , wherein the statistical measure of vertical distances is a median vertical distance.
8. A system comprising:
one or more processors; and
a memory, the memory storing instructions, which when executed cause the one or more processors to:
receive an image of a plurality of organized objects;
extract feature points from the image of the plurality of organized objects;
find matching feature points in the image of the plurality of organized objects;
cluster matching feature points with a matching distance and angle between feature points;
find multiple identical objects based on the clusters by:
finding a statistical measure of vertical distances between matching feature points in the clusters;
finding a maximum number of identical objects in a stack based on the statistical measure of vertical distances;
determining coordinates for each of the identical objects in the stack with a height corresponding to the statistical measure of vertical distances; and
determining a number of stacks of the identical objects based on matching feature points that are horizontally aligned.
9. The system of claim 8 , wherein the instructions cause the one or more processors to locate unindexed objects using groups of identical objects that don't correspond to identified indexed objects.
10. The system of claim 8 , wherein to find matching feature points in the image, the instructions cause the one or more processors to use a tree approach.
11. The system of claim 8 , wherein the instructions cause the one or more processors to:
identify feature points in a cluster that are vertically aligned;
identify feature points in the cluster that are horizontally aligned; and
identify an area surrounding the vertically and horizontally aligned feature points as an area of unindexed products.
12. The system of claim 8 , wherein the instructions cause the one or more processors to:
generate image segments based on groups of pixels in the image of the plurality of organized objects;
label the image segments with cluster information; and
wherein, to find multiple identical objects based on the clusters, the instructions further cause the one or more processors to identify image segments labeled with a cluster that have a consistent spatial relationship with other image segments labeled with the cluster.
13. The system of claim 8 , wherein to find matching feature points, the instructions cause the one or more processors to identify feature points that have a descriptor difference within a threshold.
14. The system of claim 8 , wherein the statistical measure of vertical distances is a median vertical distance.
15. A computer program product comprising a non-transitory computer usable medium including a computer readable program, wherein the computer readable program, when executed on a computer causes the computer to:
receive an image of a plurality of organized objects;
extract feature points from the image of the plurality of organized objects;
find matching feature points in the image of the plurality of organized objects;
cluster matching feature points with a matching distance and angle between feature points;
find multiple identical objects based on the clusters by:
finding a statistical measure of vertical distances between matching feature points in the clusters;
finding a maximum number of identical objects in a stack based on the statistical measure of vertical distances;
determining coordinates for each of the identical objects in the stack with a height corresponding to the statistical measure of vertical distances; and
determining a number of stacks of the identical objects based on matching feature points that are horizontally aligned.
16. The computer program product of claim 15 , wherein the computer readable program causes the computer to locate unindexed objects using groups of identical objects that don't correspond to identified indexed objects.
17. The computer program product of claim 15 , wherein the computer readable program causes the computer to:
identify feature points in a cluster that are vertically aligned;
identify feature points in the cluster that are horizontally aligned; and
identify an area surrounding the vertically and horizontally aligned feature points as an area of unindexed products.
18. The computer program product of claim 15 , wherein the computer readable program causes the computer to:
generate image segments based on groups of pixels in the image of the plurality of organized objects;
label the image segments with cluster information; and
wherein, to find multiple identical objects based on the clusters, the computer readable program causes the computer to identify image segments labeled with a cluster that have a consistent spatial relationship with other image segments labeled with the cluster.
19. The computer program product of claim 15 , wherein to find matching feature points, the computer readable program causes the computer to identify feature points that have a descriptor difference within a threshold.
20. The computer program product of claim 15 , wherein the statistical measure of vertical distances is a median vertical distance.