Method for automatically determining a region of interest for text and data overlay
View Patent ↗Provided is a method for the automatic determination of a region of interest in an image, comprising the steps of: segmenting the image into a plurality of smaller regions, each region extending over a plurality of pixels; performing an analysis on each of said regions to characterize an aspect of the region relating to its level of importance in communicating information to a viewer; grouping adjacent regions having similar aspect characteristics; and identifying at least one group as a region of interest. Also provided is a method employing the steps above for use in an automated document composition process, where overlaid content is not placed over regions of interest that are identified.
1. A method for the automatic determination of a region of interest in an image, comprising:
segmenting the image into a plurality of smaller regions, each region extending over a plurality of pixels;
performing an analysis on each of said regions to characterize an aspect of the region relating to its level of importance in communicating information to a viewer, wherein said step of performing an analysis, includes determining a contrast value as an aspect characteristic for each region and averaging the contrast values across multiple regions;
grouping adjacent regions having similar aspect characteristics; and
identifying at least one group as a region of potential overlay.
2. The method of claim 1 , wherein each region is rectangular in shape.
3. The method of claim 1 , wherein image is segmented into a predefined number of regions.
4. The method of claim 1 , wherein image is segmented in regions having at least one predefined dimension.
5. The method of claim 1 , wherein said step of averaging the contrast values across multiple regions comprises applying a filter to the regions.
6. The method of claim 5 , wherein said filter is a low-pass filter.
7. The method of claim 1 , wherein determining a contrast value for each region comprises the steps of:
for each region of the image, characterizing a local histogram signal representing the image signals within the region;
comparing each local histogram signal to a reference histogram signal; and
deriving a local contrast variance from the comparison, representing the flatness of the local histogram.
8. The method of claim 1 , wherein said step of grouping adjacent regions comprises the steps of:
identifying the contrast variance of the region with the greatest image contrast;
based upon the contrast variance of each region, classifying each of the regions, as at least a region of high relevance or a region of low relevance, as compared to a region with the greatest contrast variance; and
applying a threshold to the contrast variance values to identifying those regions with the lowest contrast.
9. The method of claim 8 , wherein said step of applying a threshold includes selecting all areas above a threshold value.
10. The method of claim 8 , wherein the step of applying a threshold includes selecting a predetermined number of regions having the highest contrast variance.
11. The method of claim 8 , wherein the step of applying a threshold includes selecting regions using a conditional threshold and determining a Boolean combination of the different averages.
12. The method of claim 1 , wherein the step of identifying at least one group as a region of potential overlay, is a function of image space required for an overlay region.
13. An automated document composition process, comprising:
receiving a document including at least one image therein;
specifying content to be overlaid on at least a portion of the image;
identifying at least one low interest region of the image upon which the content may be overlaid, including segmenting the image into a plurality of smaller regions, each region extending over a plurality of pixels, performing an analysis on each of said regions to characterize an aspect of the region relating to its level of importance in communicating information to a viewer, grouping adjacent regions having similar aspect characteristics, and identifying at least one group as a region of interest, wherein performing an analysis, comprises
determining a contrast value as an aspect characteristic for each region; and
averaging the contrast values across multiple regions; and automatically inserting at least a portion of the content into at least the one region.
14. The process of claim 13 , wherein grouping adjacent regions comprises:
identifying the contrast variance of the region with the greatest image contrast;
based upon the contrast value variance of each region, classifying each of the regions, as at least a region of high relevance or a region of low relevance, as compared to the a region with the greatest contrast variance; and
applying a threshold to the contrast variance values to identifying those regions with the lowest contrast.
15. The process of claim 13 , wherein averaging the contrast values across multiple regions comprises applying a filter to the regions.
16. The process of claim 15 , wherein said filter is a low-pass filter.
17. The process of claim 13 , wherein determining a contrast value for each region further comprises:
for each region of the image, characterizing a local histogram signal representing the image signals within the region;
comparing each local histogram signal to a reference histogram signal; and
deriving a local contrast variance from the comparison, representing the flatness of the local histogram.
18. The process of claim 13 , wherein grouping adjacent regions having similar aspect characteristics comprises:
identifying the contrast variance of the region with the greatest image contrast;
based upon the contrast variance of each region, classifying each of the regions, as at least a region of high relevance or a region of low relevance, as compared to a region with the greatest contrast variance; and
applying a threshold to the contrast variance values to identifying those regions with the lowest contrast.