1. Content-based Image Retrieval
CE 264 Xiaoguang Feng
March 14, 2002
Based on: J. Huang. Color-Spatial Image Indexing and Applications. Ph.D thesis, Cornell Univ., 1998.

2. Contents Introduction. Color-histogram vs. Correlogram.
Implementations and Results.
Conclusion.

3. Introduction Motivation of CBIR Image features for CBIR Low Level:
Color
Texture
Edge/Shape
Object Level:
Regions

4. Color Histogram The histogram of image I is defined as:
For a color Ci , Hci(I) represents the number of pixels of color Ci in image I .
OR:
For any pixel in image I, Hci(I) represents the possibility of that pixel is in color Ci.
Most commercial CBIR systems include color histogram as one of the features (e.g., QBIC of IBM).
No space information.

5. Improvement of color histogram
There are several techniques proposed to integrate spatial information with color histograms:
W.Hsu, et al., An integrated color-spatial approach to content-based image retrieval. 3rd ACM Multimedia Conf. Nov 1995.
Smith and Chang, Tools and techniques for color image retrieval, SPIE Proc. 2670, 1996.
Stricker and Dimai, Color indexing with weak spatial constraints, SPIE Proc. 2670, 1996.
Gong, et al., Image indexing and retrieval based on human perceptual color clustering, Proc. 17th IEEE Conf. On Computer Vision and Pattern Recognition, 1998.
Pass and Zabih, Histogram refinement for content-based image retrieval. IEEE Workshop on Applications of Computer Vision, 1996.
Park, et al., Models and algorithms for efficient color image indexing. Proc. Of IEEE Workshop on Content-Based Access of Image and Video Libraries, 1997.

6. Color auto-correlogram
Pick any pixel p1 of color Ci in the image I, at distance k away from p1 pick another pixel p2, what is the probability that p2 is also of color Ci?
Red ? k P2 P1
Image: I

7. Color auto-correlogram
The auto-correlogram of image I for color Ci , distance k:
Integrate both color information and space information.

8. Color auto-correlogram

9. Implementations Pixel Distance Measures
Use D8 distance (also called chessboard distance):
Choose distance k=1,3,5,7
Computation complexity:
Histogram:
Correlogram:

10. Implementations Features Distance Measures:
D( f(I1) - f(I2) ) is small  I1 and I2 are similar.
Example: f(a)=1000, f(a’)=1050; f(b)=100, f(b’)=150
For histogram:
For correlogram:

11. Test Environment
300 Color Images: flowers, people, scene, etc.

12. Test Environment Image quantized to 512 and 64 colors.
First calculate the correlogram and histogram of the 300 images, saved as data file.
For each query, calculate the correlogram and histogram of the query image; compare it with the data file; sort the feature distances.
The order of the target image in the sorted searching result measures the performance.

13. Test Results
If there is no difference between the query and the target images, both methods have good performance.
Correlogram method
Query Image
(512 colors)
1st
2nd
3rd
4th
5th
Histogram method
1st
2nd
3rd
4th
5th

14. Test Results
The correlogram method is more stable to color change than the histogram method.
Query
Correlogram method: 1st
Histogram method: 48th
Target

15. Test Results
The correlogram method is more stable to large appearance change than the histogram method.
Query
Correlogram method: 1st
Histogram method: 31th
Target

16. Test Results
The correlogram method is more stable to contrast & brightness change than the histogram method.
Query 1
Query 2
Query 3
Query 4
C: 178th
H: 230th
C: 1st
H: 1st
C: 1st
H: 3rd
C: 5th
H: 18th
Target

17. Conclusion
The color correlogram describes the global distribution of local spatial correlations of colors.
It’s easy to compute.
It’s more stable than the color histogram method.

18. Thanks