- 期刊
- OpenAccess
Color Constancy using Ridge Regression
並列摘要
Although there exist a number of single color constancy algorithms, none of them can be considered universal. Consequently, how to select and combine existing single algorithms are two important research directions in the field of color constancy. In this paper we use ridge regression, a simple yet effective machine learning approach, to select and combine existing color constancy algorithms. Two algorithms are proposed using ridge regression in this paper. In the first method (combination of existing color constancy algorithms), the color of the light source is estimated based on the results of existing single color constancy algorithms using ridge regression. In the second method (selection of existing color constancy algorithms), the proper algorithm for a given image is selected by ridge regression based on natural image statistics. Then it is used to estimate the color of the light source. The proposed two algorithms are verified on a large data set with more than 11,000 images. The experimental results demonstrate that the proposed methods are competitive with the state-of-the-art methods.
延伸閱讀
- 郭晨渝（2015）。A Novel Effective and Efficient Color Image Compression Technique Based on Color Replacement〔碩士論文，國立屏東科技大學〕。華藝線上圖書館。https://doi.org/10.6346/NPUST.2015.00106
- 楊怡群（2022）。Research on Automated Deep Color Grading Based on Color Taste〔碩士論文，國立臺灣師範大學〕。華藝線上圖書館。https://doi.org/10.6345/NTNU202200269
- Chien, W. C. (2017). Depth Image Recovery Based on Correlative Color Image Structure and Feature Matching [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU201700139
- 翁秉義（2009）。Realism Assessment of Color Compatibility and Re-coloring Using a Single Image〔碩士論文，國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU.2009.00409
- 陳君彥（2004）。Quality Assessment and Preference Color Processing for Color Images。光學工程，(88)，21-30。https://doi.org/10.30011/OE.200412.0002