¹¦·ò:   2013Äê6ÔÂ28ÈÕ£¨ÖÜÎ壩ÉÏÎç9:00

µØÖ·: ±¦É½Ð£ÇøEÂ¥6²ãÎ÷×ù

Ñݽ²Õß:  Ïã¸Û¿Æ¼¼´óѧ Oscar Chi-Lim Au ½ÌÊÚ

½²×ùÌáÒª£º

Image colorization, which is the process of adding color to a monochrome image, used to be a timeconsuming and tedious task that requires tremendous user efforts. Traditionally, image colorization involves segmenting the given grayscale image into regions, then the user proceeds to assign a color to each region. Although it is expensive, colorization is widely used for coloring black-and-white photos and image recoloring. Existing colorization algorithms receive color cues in form of either scribbles that indicate colors of certain pixels or images with similar color; while their mechanisms to propagate chrominance information vary a lot.

In this work, we present a novel method to perform image colorization using sparse representation. We first trains an over-complete dictionary in YUV color space. Then taking a grayscale image and a subset of color pixels as inputs, our algorithm colorizes overlapping image patches via sparse representation; it is achieved by seeking sparse representations of patches that are consistent with both the grayscale image and the color pixels. We then aggregate the colorized patches with weights to obtain the colorization result. Our method leads to high-quality colorizations with small number of given color pixels. To demonstrate one of the applications of the proposed method, we apply it to transfer the color of one image onto another to obtain a visually pleasing image.

Ó­½Ó¿í´óÀÏʦºÍѧÉú²ÎÓ룡