Systems and methods for organizing image data into regions
First Claim
1. A method for organizing regions identified in image data, comprising:
- averaging colors of regions of a spatial cluster to generate an average color;
assigning the average color as a color of the spatial cluster;
setting a size of the spatial cluster to a total number of pixels contained in the spatial cluster;
forming one or more color clusters, each color cluster including regions of the image data having a color difference that is less than a color threshold;
grouping two or more regions included in a color cluster into one of one or more spatial clusters if closest boundaries of the two or more regions are within a distance threshold;
assigning each spatial cluster to a binary output plane;
and eliminating regions from color or spatial clusters which are determined to be completely contained within other regions.
5 Assignments
0 Petitions
Accused Products
Abstract
The image data is initially processed in a manner that is similar to that used to convert the image data into 3-layer mixed raster content format. After the image data is initially processed through a first stage of the three-layer segmentation process, rather than completing the segmentation into foreground and background planes, the image data is analyzed to identify regions having similar image characteristics. These regions are grouped together in cluster if they share color characteristics and are spatially close. The clusters are then sorted according to their size, and placed into a binary foreground plane based on one or more color characteristics. Each cluster gathered into a binary foreground plane having the same color as the color of the cluster.
67 Citations
16 Claims
-
1. A method for organizing regions identified in image data, comprising:
-
averaging colors of regions of a spatial cluster to generate an average color; assigning the average color as a color of the spatial cluster; setting a size of the spatial cluster to a total number of pixels contained in the spatial cluster; forming one or more color clusters, each color cluster including regions of the image data having a color difference that is less than a color threshold; grouping two or more regions included in a color cluster into one of one or more spatial clusters if closest boundaries of the two or more regions are within a distance threshold; assigning each spatial cluster to a binary output plane; and eliminating regions from color or spatial clusters which are determined to be completely contained within other regions. - View Dependent Claims (2, 3, 14, 15, 16)
-
-
4. A method for organizing regions identified in image data, comprising:
-
dividing an image area of the image data into a plurality of tiles; assigning one or more regions of the image data to a tile based on a location of a bounding box of each of the regions; forming one or more color clusters by including, in each color cluster, regions of the tile that have colors that differ by less than a color threshold, or grouping two or more regions included in a first color cluster into a spatial cluster, if closest boundaries of the two or more regions are within a distance threshold; creating a new color cluster for a first region within the tile if the first region is not included in any other color cluster; creating a new spatial cluster within the new color cluster if a closest boundary of a region of the new color cluster not included in an existing spatial cluster within the new color cluster is greater than a distance threshold away from a boundary of any other region of the existing spatial cluster within the new color cluster; and a) combining 2×
2 blocks of tiles into new tiles;b) combining color clusters within a new tile if colors of regions of the color clusters within the new tile are less than the color threshold apart; and combining spatial clusters within the new tile if boundaries of regions of the spatial clusters within the new tile are less than the distance threshold apart; c) renaming the new tiles as tiles; and repeating a)-c) until a single tile includes all the regions of the image data. - View Dependent Claims (5, 6, 7)
-
-
8. An apparatus for processing regions of image data, comprising:
-
a color cluster processor that forms one or more color clusters by grouping regions of the image data which are within a color threshold of each other, and/or a spatial cluster processor that forms one or more spatial clusters for each of the color clusters, regions of the color cluster being included in the spatial cluster when their respective bounding boxes are within a distance threshold of each other; and a planes generator which creates binary output planes based on the color or spatial clusters and an inner blob module which eliminates regions from color and/or spatial clusters which are determined to be completely contained within other regions; wherein the color cluster processor also divides an image area of image data into a set of tiles, and assigns the regions of the image data to the tiles based on a location of a bounding box of each of the regions, and groups the regions into clusters depending on color and/or spatial characteristics of the regions. - View Dependent Claims (9, 10, 11, 12, 13)
-
Specification