Method and system for describing image region based on color histogram

Patent metadata

Field	Value
Publication number	US-9275447-B2
Application number	US-201213671240-A
Country	US
Kind code	B2
Filing date	Nov 7, 2012
Priority date	Nov 9, 2011
Publication date	Mar 1, 2016
Grant date	Mar 1, 2016

Abstract

A method and system for describing image region based on color histogram is provided. The method for describing an image region based on color histogram comprising: a color quantization step for quantizing said image region into a quantized color region, wherein a specific color range and colors outside the specific color range are differently treated in the quantizing; a color histogram calculation step for computing one or more color histograms according to said quantized color region; and a histogram assembling step for generating a unique vector by using said one or more color histograms.

First claim

What is claimed is:

1. A method for describing an image region based on color histogram, comprising:
transforming the image region into HSV or HSL color space;
extracting a specific color range, which includes colors perceptible by human vision as black, gray and white colors, from the HSV or HSL color space;
dividing the HSV or HSL color space except for the specific color range into a plurality of common color ranges;
quantizing the image region into a quantized color region by quantizing the specific color range according to a Value plane of the HSV color space or a Lightness plane of the HSL color space and quantizing each of the common color ranges according to the Value plane and a Saturation plane of the HSV color space or the Lightness plane and a Saturation plane of the HSL color space;
computing one or more color histograms according to the quantized color region;
smoothing histogram bins corresponding to the common color ranges except for the specific color range in the one or more color histograms so that adjacent histogram bins have an impact on each other; and
generating a unique vector by using the one or more color histograms.

2. The method according to claim 1 , further comprising dividing the quantized color region into a plurality of image blocks,
wherein in the computing step, a color histogram is computed for each of the plurality of image blocks, and
wherein in the generating step, the unique vector is generated by using the computed color histograms.

3. The method according to claim 1 , further comprising dividing the image region into a plurality of image blocks,
wherein in the quantizing step, the quantization is performed on each of the plurality of image blocks, to form a quantized color region containing a plurality of image blocks,
wherein in the computing step, a color histogram is computed for each of the plurality of image blocks, and
wherein in the generating step, the unique vector is generated by using the computed color histograms.

4. The method according to claim 2 , wherein some of the plurality of image blocks are overlapped.

5. The method according to claim 2 , wherein the plurality of image blocks are ring-shaped, have the same central point and have different radiuses.

6. The method according to claim 2 , wherein the plurality of image blocks are grid-shaped or radial-shaped.

7. The method according to claim 2 , wherein the computing step comprises
computing a histogram for each image block in the quantized color region, and wherein the method further comprises
normalizing the the histogram in each image block as a final color histogram.

8. A system for describing an image region based on color histogram, comprising:
a transformation unit configured to transform the image region into HSV or HSL color space;
an extraction unit configured to extract a specific color range, which includes colors perceptible by human vision as black, gray and white colors, from the HSV or HSL color space;
a dividing unit configured to divide the HSV or HSL color space except for the specific color range into a plurality of common color ranges;
a color quantization unit configured to quantize the image region into a quantized color region by quantizing the specific color range according to a Value plane of the HSV color space or a Lightness plane of the HSL color space and quantizing each of the common color ranges according to the Value plane and a Saturation plane of the HSV color space or the Lightness plane and a Saturation plane of the HSL color space;
a color histogram calculation unit configured to compute one or more color histograms according to the quantized color region;
a smoothing unit configured to smooth histogram bins corresponding to the common color ranges except for the specific color range in the one or more color histograms so that adjacent histogram bins have an impact on each other; and
a histogram assembling unit configured to generate a unique vector by using the one or more color histograms.

9. The system according to claim 8 , further comprising a block dividing unit configured to divide the quantized color region into a plurality of image blocks,
wherein the color histogram calculation unit is configured to compute a color histogram for each of the plurality of image blocks, and
wherein the histogram assembling unit is configured to generate the unique vector by using the computed color histograms.

10. The system according to claim 8 , further comprising a block dividing unit configured to divide the image region into a plurality of image blocks,
wherein the color quantization unit is configured to perform the quantization on each of the plurality of image blocks, to form a quantized color region containing a plurality of image blocks,
wherein the color histogram calculation unit is configured to compute a color histogram for each of the plurality of image blocks, and
wherein the histogram assembling unit is configured to generate the unique vector by using the computed color histograms.

11. The system according to claim 9 , wherein some of the plurality of image blocks are overlapped.

12. The system according to claim 9 , wherein the plurality of image blocks are ring-shaped, have the same central point and have different radiuses.

13. The system according to claim 9 , wherein the plurality of image blocks are grid-shaped or radial-shaped.

14. The system according to claim 9 , wherein the color histogram calculation unit comprises:
a unit configured to compute a histogram for each image block in the quantized color region;
a unit configured to smooth the histogram bins from all second color ranges except the specific color range; and
a unit configured to normalize the histogram in each image block as a final color histogram.

Assignees

• Canon Kk

Inventors

• Jiang Yong

Classifications

• G06T5/40 Primary

  using histogram techniques · CPC title

• G06V10/56 Primary

  relating to colour · CPC title

• G06F17/3025

  Physics · mapped topic

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title

• G06K9/4652

  Physics · mapped topic