| Presentation | 2012-11-08 Auto-Colorization of Monochrome Images Using Object Recognition and Markov Random Field Hitoshi MATSUO, Takaki MAKINO, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | We introduce a new auto-colorization method which appends color information to a monochrome picture based on a training set of various color images using specific object recognition technique. The conventional color-transferring method was based on the global information of a single color picture (e.g.a color-histogram), which had some problems like 1) it is unable to append color to proper region, 2) it requires the color data whose composition exactly matches the original monochrome picture, and 3) it has little ways to correct the wrongly-colored image. Our new auto-colorization method does not require the fully-matched picture of the whole image, because it 1) calculates SIFT features of the feature points from a training set of color images, 2) finds the match between the SIFT features of a training set and those of the monochrome picture, 3) transfers the color information at the matched features and thus 4) generates the partial coloring information. Further, we applied the Markov random field model to the color information of the monochrome picture, by which each unit corresponds to a region of the monochrome image obtained by a graph-based image segmentation method. In this way we can append color to proper region, complementing and adjusting the color information at one time based on the results by the specific object recognition technique. When a training set of color images is in the same category as that of the original monochrome picture, such as a part of movie, the resulting colored images were sufficiently consistent. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Auto-Colorization / Object Recognition / Markov Random Field |
| Paper # | IBISML2012-83 |
| Date of Issue |
| Conference Information | |
| Committee | IBISML |
|---|---|
| Conference Date | 2012/10/31(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Information-Based Induction Sciences and Machine Learning (IBISML) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Auto-Colorization of Monochrome Images Using Object Recognition and Markov Random Field |
| Sub Title (in English) | |
| Keyword(1) | Auto-Colorization |
| Keyword(2) | Object Recognition |
| Keyword(3) | Markov Random Field |
| 1st Author's Name | Hitoshi MATSUO |
| 1st Author's Affiliation | College of Arts and Sciences, the University of Tokyo:Office for the Next Generation (UROP), Institute of Industrial Science, the University of Tokyo() |
| 2nd Author's Name | Takaki MAKINO |
| 2nd Author's Affiliation | Institute of Industrial Science, the University of Tokyo |
| Date | 2012-11-08 |
| Paper # | IBISML2012-83 |
| Volume (vol) | vol.112 |
| Number (no) | 279 |
| Page | pp.pp.- |
| #Pages | 8 |
| Date of Issue | |