Color constancy using color edge moments and regularized regression in anchored neighborhood

To improve the accuracy of illumination estimation while maintaining a relative fast execution speed, a novel learning-based color constancy using color edge moments and regularized regression in an anchored neighborhood is proposed. First, scene images are represented by the color edge moments of various orders. Then, an iterative regression with a squared Frobenius norm（F-norm） regularizer is introduced to learn the mapping between the edge moments and illuminants in the neighborhood of the anchored sample.Illumination estimation for the test image finally becomes the nearest anchored point search followed by a matrix multiplication using the associated mapping matrix which can be precalculated and stored. Experiments on two standard image datasets show that the proposed approach significantly outperforms the state-of-the-art algorithms with a performance increase of at least 10. 35% and 7. 44% with regard to median angular error.

RESEARCH ON COLOR CONSTANCY UNDER OPEN ILLUMINATION CONDITIONS

Color information plays a key role in the research fields of object recognition and image retrieval. However, the actual color varies by the conditions of illumination, especially the open natural daylight. Four different color constancy schemes are proposed in the paper to minimize the effects of open illumination conditions. (1) The color constancy scheme based on the image statistics is proposed, which includes the color cast detection and removal. (2) The color constancy scheme based on the color temperature curve is proposed, which combines Gaussian model with linear fitting to estimate color temperature curve. (3) The color constancy scheme based on the double exposure theory is proposed, which is able to reproduce a color image under typical illumination. (4) According to the concepts of supervised learning, the supervised color constancy scheme is proposed. The transformation of color values from unknown illumination to typical illumination is solved by improved Support Vector Regression (SVR).

A New Color Constancy Model for Machine Vision

Both physiological and psychological evidences suggest that the human visual system analyze images in neural subsystems tuned to different attributes of the stiamlus. Color module and lightness module are such subsystems. Under this general result, a new physical model of trichromatic system has been developed to deal with the color constancy of computer vision. A normal color image is split into two images: the gray scale image and the equal lightness color image for the two modules. Relatively a two-dimensional descriptor is applied to describe the property of surface reflectance in the equal lightness color image. This descrip- tion of surface spectral reflectance has the property of color constancy Image segmentation experiments based on color property of object show that the presented model is effective.

Estimation of illumination chromaticity via adaptive reduced relevance vector machine

A new regression algorithm of an adaptive reduced relevance vector machine is proposed to estimate the illumination chromaticity of an image for the purpose of color constancy. Within the framework of sparse Bayesian learning, the algorithm extends the relevance vector machine by combining global and local kernels adaptively in the form of multiple kernels, and the improved locality preserving projection （LLP） is then applied to reduce the column dimension of the multiple kernel input matrix to achieve less training time. To estimate the illumination chromaticity, the algorithm is trained by fuzzy central values of chromaticity histograms of a set of images and the corresponding illuminants. Experiments with real images indicate that the proposed algorithm performs better than the support vector machine and the relevance vector machine while requiring less training time than the relevance vector machine.

AN INVESTIGATION OF RETINEX ALGORITHMS FOR IMAGE ENHANCEMENT

This letter introduces color constancy and Retinex theory for image enhancement.It clas- sifies Retinex algorithms into four categories and provides their principles and implementations in general.The experimental results of Frankle-McCann,MSR (Multi-Scale Retinex) and PNSD (Pro- jected Normalized Steepest Descent) Retinex algorithms are presented and compared.Moreover, variance and average gradient are proposed to evaluate the performance of the different algorithms.