Example-based image colorization via automatic feature selection and fusion

Li, Bo; Lai, Yu‐Kun; Rosin, Paul L.

doi:10.1016/j.neucom.2017.05.083

Cited by 23 publications

(9 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to eliminate halo effects near strong contours, [7] introduced a coupled regularization term with luminance channel to preserve image contours during the colorization process. Instead of using a static combination of multiple features to improve the matching performance, [26] proposed an automatic feature selection based image colorization method via a Markov Random Field (MRF) model. Our method also takes a single color image as reference.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Example-Based Image Colorization Using Locality Consistent Sparse Representation

Zhao

et al. 2017

IEEE Trans. on Image Process.

Self Cite

View full text Add to dashboard Cite

Image colorization aims to produce a natural looking color image from a given gray-scale image, which remains a challenging problem. In this paper, we propose a novel example-based image colorization method exploiting a new locality consistent sparse representation. Given a single reference color image, our method automatically colorizes the target gray-scale image by sparse pursuit. For efficiency and robustness, our method operates at the superpixel level. We extract low-level intensity features, mid-level texture features, and high-level semantic features for each superpixel, which are then concatenated to form its descriptor. The collection of feature vectors for all the superpixels from the reference image composes the dictionary. We formulate colorization of target superpixels as a dictionary-based sparse reconstruction problem. Inspired by the observation that superpixels with similar spatial location and/or feature representation are likely to match spatially close regions from the reference image, we further introduce a locality promoting regularization term into the energy formulation, which substantially improves the matching consistency and subsequent colorization results. Target superpixels are colorized based on the chrominance information from the dominant reference superpixels. Finally, to further improve coherence while preserving sharpness, we develop a new edge-preserving filter for chrominance channels with the guidance from the target gray-scale image. To the best of our knowledge, this is the first work on sparse pursuit image colorization from single reference images. Experimental results demonstrate that our colorization method outperforms the state-of-the-art methods, both visually and quantitatively using a user study.

show abstract

Section: Related Workmentioning

confidence: 99%

“…Guided image filtering [46] is well known for improved structure preservation by using a guidance image and a locally adaptive linear model. We use the target grayscale image to guide the filtering of the chrominance images as [26] does. Fig.…”

Section: Edge-preserving Joint Filtering Guided By Luminancementioning

confidence: 99%

Example-Based Image Colorization Using Locality Consistent Sparse Representation

Zhao

et al. 2017

IEEE Trans. on Image Process.

Self Cite

View full text Add to dashboard Cite

show abstract

“…It is also problematic that the same intensity may represent different colors, so there is no exact solution [8]. In general, existing colorization methods can be divided into three main categories, all of which have limitations: user-scribbled methods, example-based methods, and those that employ a large number of training images [9]. User-scribbled techniques [10][11][12][13] require a user to manually add colored marks to a grayscale image [13].…”

Section: Introductionmentioning

confidence: 99%

“…Moreover, it is almost impossible to add such markings to large volumes (gigabytes) of aerial imagery. In the case of the example-based method [1,8,9,[14][15][16], it typically transfers the color information from a similar reference image to the input grayscale image rather than obtaining chromatic values from the user, thereby reducing the burden on users. However, as feature matching is critical to the quality of the results, satisfactory results cannot be obtained if feature matching is not performed correctly [15].…”

Section: Introductionmentioning

confidence: 99%

Learning-Based Colorization of Grayscale Aerial Images Using Random Forest Regression

Seo

Kim

et al. 2018

Applied Sciences

View full text Add to dashboard Cite

Image colorization assigns colors to a grayscale image, which is an important yet difficult image-processing task encountered in various applications. In particular, grayscale aerial image colorization is a poorly posed problem that is affected by the sun elevation angle, seasons, sensor parameters, etc. Furthermore, since different colors may have the same intensity, it is difficult to solve this problem using traditional methods. This study proposes a novel method for the colorization of grayscale aerial images using random forest (RF) regression. The algorithm uses one grayscale image for input and one-color image for reference, both of which have similar seasonal features at the same location. The reference color image is then converted from the Red-Green-Blue (RGB) color space to the CIE L*a*b (Lab) color space in which the luminance is used to extract training pixels; this is done by performing change detection with the input grayscale image, and color information is used to establish color relationships. The proposed method directly establishes color relationships between features of the input grayscale image and color information of the reference color image based on the corresponding training pixels. The experimental results show that the proposed method outperforms several state-of-the-art algorithms in terms of both visual inspection and quantitative evaluation.

show abstract

“…They begin by dividing the grayscale image into non-overlapping windows, and then, for each window, two pixels that convergent the average luminance of window are chosen as the seeds, Then the seed is colored by the user, where they used optimization reduces the variation between the seeds and their neighbouring pixels that used to deploy the colors to the other pixels. Li et al [11] proposed method for coloring gray images by using the automatic outcome-finding feature with outcomes combined through Markov Random Field (MRF) model to get better coloring.…”

mentioning

confidence: 99%

Automatic gray images colorization based on lab color space

Abbadi

Razaq

2020

IJEECS

View full text Add to dashboard Cite

<p>The colorization aim to transform a black and white image to a color image. This is a very hard issue and usually requiring manual intervention by the user to produce high-quality images free of artifact. The public problem of inserting gradients color to a gray image has no accurate method. The proposed method is fully automatic method. We suggested to use reference color image to help transfer colors from reference image to gray image. The reference image converted to Lab color space, while the gray scale image normalized according to the lightness channel L. the gray image concatenate with both a, and b channels before converting to RGB image. The results were promised compared with other methods.</p>

show abstract

Example-based image colorization via automatic feature selection and fusion

Cited by 23 publications

References 27 publications

Example-Based Image Colorization Using Locality Consistent Sparse Representation

Example-Based Image Colorization Using Locality Consistent Sparse Representation

Learning-Based Colorization of Grayscale Aerial Images Using Random Forest Regression

Automatic gray images colorization based on lab color space

Contact Info

Product

Resources

About