Guided batching
The present invention provides a method of generating a robust global map using a plurality of limited field-of-view cameras to capture an environment. Provided is a method for generating a three-dimensional map comprising: receiving a plurality of sequential image data wherein each of the plurality of sequential image data comprises a plurality of sequential images, further wherein the plurality of sequential images is obtained by a plurality of limited field-of-view image sensors; determining a pose of each of the plurality of sequential images of each of the plurality of sequential image data; determining one or more overlapping poses using the determined poses of the sequential image data; selecting at least one set of images from the plurality of sequential images wherein each set of images are determined to have overlapping poses; and constructing one or more map portions derived from each of the at least one set of images.
1. A computer-implemented method comprising:
determining, by a computing system, subsets of image data associated with an area, wherein the subsets of image data are based on a set of images captured at the area;
determining, by the computing system, a first group of the subsets of image data having image properties that are separated by a first distance that exceeds a similarity distance threshold to each other;
determining, by the computing system, a second group of the subsets of image data having image properties that are separated by a second distance that is within the similarity distance threshold to each other;
excluding, by the computing system, the first group of the subsets of image data from the determined subsets of image data based on the first distance exceeding the similarity distance threshold; and
generating, by the computing system, a map portion associated with the area based on the excluded first group of the subsets of image data and the second group of the subsets of image data.
2. The method of claim 1 , further comprising:
generating, by the computing system, a graph of the subsets of image data, wherein nodes in the graph correspond to the subsets of image data, edges in the graph connect nodes that are within the similarity distance threshold, and the edges are weighted based on the image properties for the subsets of image data corresponding to the nodes connected by the edges.
3. The method of claim 2 , further comprising:
partitioning, by the computing system, the graph into subgraphs based on a measure of dissimilarity between the image data of the subgraphs and a measure of similarity between the image data within the subgraphs.
4. The method of claim 3 , wherein the partitioning the graph into subgraphs comprises performing a first level graph cut based on the edges that are weighted below a minimum weighting threshold.
5. The method of claim 4 , wherein the partitioning the graph further comprises performing a second level graph cut based on visual similarity or orientation similarity within the subgraphs.
6. The method of claim 5 , wherein the partitioning the graph further comprises recursively performing further graph cuts until each subgraph is within a threshold number of edges and a threshold number of nodes.
7. The method of claim 1 , further comprising:
generating, by the computing system, an alignment set of image data by overlapping at least one subset of the subsets of image data with at least one other subset of the subsets of image data, wherein corresponding images in the overlapping subsets are used as connection points in generating a map of the area.
8. The method of claim 1 , further comprising:
excluding, by the computing system, image data from the second group of the subsets of image data based on a success metric associated with a structure from motion (SFM) process.
9. The method of claim 1 , wherein the subsets of image data include at least one straight subset and at least one turn subset, wherein the at least one straight subset includes images associated with a dominant pose and the at least one turn subset includes images associated with two or more dominant poses.
10. The method of claim 1 , wherein the image properties include at least one of: locations, orientations, positions, poses, or visual matches associated with the set of images.
11. A system comprising:
at least one processor; and
a memory storing instructions that, when executed by the at least one processor, cause the system to perform:
determining subsets of image data associated with an area, wherein the subsets of image data are based on a set of images captured at the area;
determining a first group of the subsets of image data having image properties that are separated by a first distance that exceeds a similarity distance threshold to each other;
determining a second group of the subsets of image data having image properties that are separated by a second distance that is within the similarity distance threshold to each other;
excluding the first group of the subsets of image data from the determined subsets of image data based on the first distance exceeding the similarity distance threshold; and
generating a map portion associated with the area based on the excluded first group of the subsets of image data and the second group of the subsets of image data.
12. The system of claim 11 , wherein the instructions further cause the system to perform:
generating a graph of the subsets of image data, wherein nodes in the graph correspond to the subsets of image data, edges in the graph connect nodes that are within the similarity distance threshold, and the edges are weighted based on the image properties for the subsets of image data corresponding to the nodes connected by the edges.
13. The system of claim 12 , wherein the instructions further cause the system to perform:
partitioning the graph into subgraphs based on a measure of dissimilarity between the image data of the subgraphs and a measure of similarity between the image data within the subgraphs.
14. The system of claim 13 , wherein the partitioning the graph into subgraphs comprises performing a first level graph cut based on the edges that are weighted below a minimum weighting threshold.
15. The system of claim 14 , wherein the partitioning the graph further comprises performing a second level graph cut based on visual similarity or orientation similarity within the subgraphs.
16. A non-transitory computer-readable storage medium including instructions that, when executed by at least on processor of a computing system, cause the computing system to perform:
determining subsets of image data associated with an area, wherein the subsets of image data are based on a set of images captured at the area;
determining a first group of the subsets of image data having image properties that are separated by a first distance that exceeds a similarity distance threshold to each other;
determining a second group of the subsets of image data having image properties that are separated by a second distance that is within the similarity distance threshold to each other;
excluding the first group of the subsets of image data from the determined subsets of image data based on the first distance exceeding the similarity distance threshold; and
generating a map portion associated with the area based on the excluded first group of the subsets of image data and the second group of the subsets of image data.
17. The non-transitory computer-readable storage medium of claim 16 , wherein the instructions further cause the computing system to perform:
generating a graph of the subsets of image data, wherein nodes in the graph correspond to the subsets of image data, edges in the graph connect nodes that are within the similarity distance threshold, and the edges are weighted based on the image properties for the subsets of image data corresponding to the nodes connected by the edges.
18. The non-transitory computer-readable storage medium of claim 17 , wherein the instructions further cause the computing system to perform:
partitioning the graph into subgraphs based on a measure of dissimilarity between the image data of the subgraphs and a measure of similarity between the image data within the subgraphs.
19. The non-transitory computer-readable storage medium of claim 18 , wherein the partitioning the graph into subgraphs comprises performing a first level graph cut based on the edges that are weighted below a minimum weighting threshold.
20. The non-transitory computer-readable storage medium of claim 19 , wherein the partitioning the graph further comprises performing a second level graph cut based on visual similarity or orientation similarity within the subgraphs.