1 / 45

An Multiple Regression Analysis Based Color Transform Between Objects

An Multiple Regression Analysis Based Color Transform Between Objects. Speaker : Chen-Chung Liu. Outline. Introduction The proposed algorithm Color Objects Extraction Algorithm Using Multiple Thresholds (COEMT) Color Transform Using Multiple Regression Analysis (MRA) Conclusions.

ince
Download Presentation

An Multiple Regression Analysis Based Color Transform Between Objects

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. An Multiple Regression Analysis Based Color Transform Between Objects Speaker:Chen-Chung Liu

  2. Outline • Introduction • The proposed algorithm • Color Objects Extraction Algorithm Using Multiple Thresholds (COEMT) • Color Transform Using Multiple Regression Analysis (MRA) • Conclusions

  3. 1. Introduction(1/3) • Art purpose

  4. 1. Introduction (2/3) • Image analysis (details increasing)

  5. 1. Introduction (3/3) • Image analysis (image simplify)

  6. The proposed algorithm Figure 1. The flow chart of the proposed color transformation algorithm.

  7. 2.1.Color Objects Extraction(1/17) Figure 2. Color objects extraction algorithm flow chart.

  8. Figure 3. Pixels values distribution on different planes. 2.1.Color Objects Extraction (2/17)

  9. 2.1.Color Objects Extraction (3/17) Figure 4. Intensity versus RGB and saturation versus RGB.

  10. 2.1.Color Objects Extraction (4/17) Figure 5. The flow chart of EOAFF on HSI domain.

  11. 2.1.Color Objects Extraction(5/17)

  12. 2.1.Color Objects Extraction(6/17)

  13. 2.1.Color Objects Extraction(7/17)

  14. 2.1.Color Objects Extraction(8/17)

  15. 2.1.Color Objects Extraction(9/17) • Filter’s thresholds of hue , saturation , and intensity

  16. 2.1.Color Objects Extraction(10/17) Figure 6. An example of the proposed adaptive forecasting filter‘s working.

  17. 2.1.Color Objects Extraction(11/17) union result original image CS result BSE result Figure 7. An example of the proposed scheme.

  18. 2.1.Color Objects Extraction(12/17) original imagewith seeds DTS in RGB DTS in HSI proposed scheme Figure 8. Test image: Pink hat. C. C. Liu and G. N. Hu, Color Objects Extraction Scheme Using Dynamic Thresholds (DTS), 2009 Workshop on Consumer Electronics (WCE2009), pp. 1130-1138, 2009.

  19. 2.1.Color Objects Extraction(13/17) original imagewith seeds DTS in RGB DTS in HSI proposed scheme Figure 9. Test image: Flowers.

  20. 2.1.Color Objects Extraction(14/17) original image with seeds DTS in RGB DTS in HSI proposed scheme Figure 10. Test image: Pottery.

  21. 2.1.Color Objects Extraction(15/17) original image with seeds DTS in RGB DTS in HSI proposed scheme Figure 11. Test image: Cup set.

  22. 2.1.Color Objects Extraction(16/17) DTS in RGB original image with seeds DTS in HSI proposed scheme Figure 12. Test image: Sun flower.

  23. 2.1.Color Objects Extraction(17/17) Table 1. Comparisons of extraction results

  24. 2.2. MRA_based Color Transform(1/20) Multiple Regression Analysis (1/5) For data of ordered pairs We want to predict y from x by finding a function that fits the data as closely as possible.

  25. 2.2. MRA_based Color Transform(2/20) Multiple Regression Analysis (2/5) • MRA is used to find a polynomial function of degree , as the predicting function, that has the minimum of the sum of squares of the errors(SSE) between the predicted values of y and the observed values for all of the n data points .

  26. 2.2. MRA_based Color Transform(3/20) Multiple Regression Analysis (3/5) • The values of , , ,…,and that minimizeare obtained by setting the first partial derivatives , ,…, andequal to zero.

  27. 2.2. MRA_based Color Transform(4/20) Multiple Regression Analysis (4/5) • Solving the resulting simultaneous linear system of the so-called normal equations:

  28. 2.2. MRA_based Color Transform(5/20) Multiple Regression Analysis (5/5) • The matrix form solution bewhere

  29. 2.2. MRA_based Color Transform(6/20) Figure 13. Target object .

  30. 2.2. MRA_based Color Transform(7/20) Figure 14. Source object .

  31. 2.2. MRA_based Color Transform(8/20) • Best fitting functions Red Green Blue Figure 15. The curves of degree1, 5, and 9 best fitting functions.

  32. 2.2. MRA_based Color Transform(9/20) Figure 16. The color transfer results corresponding to the variation in the degree of best fitting polynomials.

  33. 2.2. MRA_based Color Transform(10/20) L* a* b* Figure 17. The box-plots of L*, a*, and b* for the target, source, and color transferred objects in Figure 11.

  34. 2.2. MRA_based Color Transform(11/20) Table 2. The measurement metrics for the target, source and color transferred objects in Figure 17 (1/2)

  35. 2.2. MRA_based Color Transform(12/20) Table 2. The measurement metrics for the target, source and color transferred objects in Figure 17 (2/2)

  36. 2.2. MRA_based Color Transform(13/20) • The target RGB color image is a girl in a blue dress (350×350 pixels). • The source color images with different sizes.

  37. 2.2. MRA_based Color Transform(14/20) • The extraction procedure lasted between 3 and 25 seconds, and the color transferring procedure lasted about 0.03 seconds.

  38. 2.2. MRA_based Color Transform(15/20) Figure 20. Examples of color transferring between objects with the proposed multiple regression analysis algorithm (1/2).

  39. 2.2. MRA_based Color Transform(16/20) Figure 21. Examples of color transferring between objects with the proposed multiple regression analysis algorithm (2/2).

  40. 2.2. MRA_based Color Transform(17/20) • Performance measures function:

  41. 2.2. MRA_based Color Transform(18/20) Table 3. The measurement metrics for the target and source objects in Figures 20,21

  42. 2.2. MRA_based Color Transform(19/20) Table 4. The measurement metrics for the color transferred target objects in Figures 20, 21

  43. 2.2. MRA_based Color Transform(20/20) Table 5. The absolute difference in measurement metrics of the transferred target-object from the source object in Figures 20 and 21

  44. Conclusions • Simple, effective and accurate in color transferring between objects. • Details of target object can be changed by the color complexity of source object. • Time consumption is independent of the number of bins selected and the degree of regression. • Dynamic ranges of colors of objects don’t have any restriction.

  45. Thank You Questions and Comments

More Related