Image Color Correction Based On Retinex Theory

As one of the significant media which carries and transmits the information forhuman-beings, color digital images are tend to be influenced by illumination duringacquisition, which results in the phenomenon of color cast, and thus decline the qualityof images. How to effectively suppress the influence of illumination on the imagequality, and to provide more realistic and reliable image color information forsubsequent processing, has always been considered as one of the key areas and hotissues in the research field of color digital image processing.For the halos, noise amplification and desaturation problems presented in theoutputs of existing color correction algorithm based on the Retinex theory, with addinga constraint of illumination information estimation, this paper proposed acenter/surround Retinex color correction algorithm based on bilateral filtering.Extensive simulation results showed that compared to the existing color correctionalgorithms, by enhancing the saturation and suppressing the influence of the halos andnoise to the quality of the images, the proposed algorithm can improve the subjectivequality of color digital image efficiently.For the color digital image captured under low light condition, it is difficult to meetthe needs of human beings by correct the color of the low-light image with the-existingRetinex based color correction algorithm. Based on the analyzing of the characteristicsof the low-light images, a time-spatial domain Retinex color correction algorithm basedon white balance theory is proposed in this paper to solve the problem mentioned above.Extensive simulation results showed that compared to the existing color correctionalgorithm, by correcting the color of the low-light images and suppressing the noise inthe low-light images, the subjective quality of the color digital image can be improvedefficiently.