Method and apparatus for determining regions of interest in images and for image transmission

• US 20040042656A1
• Filed: 04/08/2003
• Published: 03/04/2004
• Est. Priority Date: 10/09/2000
• Status: Active Grant

• Current Assignee: Motorola Solutions, Inc.
• Original Assignee: Motorola, Inc. (Motorola Solutions, Inc.)
• Inventors: Hobson, Paola Marcella, Timor, Kadir
• Granted Patent: US 7,162,081 B2
• Time in Patent Office: Days
• US Class Current: 382/171
• CPC Class Codes: G06T 2207/20132 Image cropping
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• CPC Class Codes: G06V 10/25 Determination of region of interest [ROI] or a volume of interest [VOI]
• CPC Class Codes: G06V 10/462 Salient features, e.g. scale invariant feature transforms [SIFT]
• CPC Class Codes: G06V 10/507 Summing image-intensity values; Histogram projection analysis
• CPC Class Codes: H04N 19/00 Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
• CPC Class Codes: H04N 19/17 the unit being an image region, e.g. an object

Abstract

A method and apparatus for ranking pixels in an image by degrees of saliency, including: defining (200, 210) a plurality of regions for each of a plurality of pixels in an image, where each of the regions for any one of the pixels at coordinates (x,y) has a different scale s, calculating (230) an thropy value for each of the regions, identifying (260) any of the regions as having a peak entropy value, wighting (270) the peak entropy value of each of the peak entropy regions by a weight value corresponding to a peak width esimate of the peak entropy value, and randking (290) the regions by the weighted peak entropy values, thereby ranking the pixels. The invention can be included in a method and apparatus for image transmission, providing preferential transmission of salient portions of the image. This is particularly advantageous in image transmission over radio links of limited bandwith.

Claims

1. A method for ranking pixels in an image by degrees of saliency, the method comprising the steps of:

a) defining (200, 210) a plurality of regions for each of a plurality of pixels in an image, wherein each of said regions, for any one of said pixels at coordinates (x,y), has a different scale s;

b) calculating (230) an entropy value for each of said regions;

c) identifying (260) any peak entropy regions, the peak entropy regions being regions which include a peak entropy value;

d) weighting (270) said peak entropy value of each of said peak entropy regions by a weight value corresponding to a peak width estimate of said peak entropy value; and

e) ranking (290) said regions by said weighted peak entropy values, thereby ranking said pixels.

2. A method according to claim 1 wherein said calculating step b) comprises calculating (220) a probability density function (PDF) estimate for said pixels within each of said regions.

3. A method according to claim 2 wherein said calculating step b) comprises calculating (220) using a Parzen window PDF estimator.

4. A method according to claim 2 wherein said calculating step b) comprises calculating (220) using histogram binning.

5. A method according to claim 1 and further comprising the step of applying a 3-tap averaging to said peak width estimate prior to performing said weighting step d).

6. A method according to claim 1 wherein said weighting step d) comprises weighting in accordance with a value W(x,y,s) according to the formula S(x,y,s)=H(x,y,s)·

W(x,y,s)

where

11. A method of image transmission, comprising a method according to any previous claim.

13. An image transmission device, comprising a device according to claim 11 for selecting salient regions of an image for preferential transmission.

7. A method for determining regions of saliency in an image, the method comprising the steps of:

a) defining (200,

     210) a plurality of regions for each of a plurality of pixels in the image, wherein each of said regions for any one of said pixels at coordinates (x,y) has a different scale s;

b) calculating (230) an entropy value for each of said regions;

c) identifying (260) any of said regions as having a peak entropy value;

d) weighting (270) said peak entropy value of each of said peak entropy regions by a weight value corresponding to a peak width estimate of said peak entropy value; and

e) for each selected one of said peak entropy regions (300);

i) selecting (310) at least one neighboring region of peak entropy,

ii) determining (320) the mean and variance of each of x, y, and s for said selected peak entropy region and said at least one neighboring region; and

where the largest of said variances does not exceed a predetermined threshold (330);

iii) removing (350) from the set of said peak entropy regions any regions within a predetermined distance from said mean (x,y,s) point; and

iv) storing (340) said mean values, thereby designating a region of interest.

8. A method according to claim 7 and further comprising performing step e) in order of said weighted peak entropy values.

9. A method according to claim 7 or claim 8, wherein

the regions are point regions and the variance zero.

10. A method according to any of claims 7-9, wherein

there are only two regions, and the variance calculation is discarded.

12. A device (500) for ranking pixels in an image by degrees of saliency, comprising:

a) means (510) for defining a plurality of regions for each of a plurality of pixels in an image (505), each of said regions having a different scale s for any one of said pixels at coordinates x,y,

b) means (530) for calculating an entropy value for each of said regions;

c) means (560) for identifying any peak entropy regions, the peak entropy regions being regions that include a peak entropy value;

d) means (570) for weighting said peak entropy value of each of said peak entropy regions by a weight value corresponding to a peak width estimate of said peak entropy value;

e) means (590) for ranking said regions by said weighted peak entropy values, thereby ranking said pixels.