Color Image Enhancement Algorithm Based On Hue Preserving

In the past decades, the price of image sensor and hardware processing of colorimage becomes more and more cheaper, leading to wider application of color imageprocessing technology, such as, printing, visualization and internet. While in the courseof obtaining color images, due to underexposure and illumination variation, many colorimages will produce some dark areas. Some important information in these dark areascan not be able to distinguish. Therefore, color image enhancement technique is of greatimportance whether in theory or in practice. color image enhancement is one of themost attractive field in digital image processing. The theory of enhancement techniqueis revealing fuzzy details or outstanding some interested features simply, makingimages more suitable for specific applications than before proceeding. Color imageenhancement needs not only image dynamic range improvement and image detailsoutstanding but also color hue keeping unchanged. Above all, this paper focuses onresearching color image enhancement technique based on hue preserving.Firstly, the basic model of human visual system is introduced in this paper.Secondly, three different color space models are elaborated. Finally, color constantalgorithms existed and classic contrast enhancement algorithms are given. In this paper,the merits and demerits of the RGB color space and HIS color space and theirappropriate applications is analyzed. In those two color space, hue preservingsimulation adopting histogram equalization has been put forward and as a result HIScolor space has been adopted in his paper for it is more efficient preserving hue.The main research work of this paper includes:①Analyze the drawbacks of the center/surround Retinex algorithm and propose anovel Retinex algorithm based on intensity compensation. In this paper, backgroundillumination estimation use bilateral filtering instead of Gaussian filtering. Because ofbilateral filtering considers both geometric closeness and photometric similarity, theestimated illuminant image maintain a good discontinuity and can avoid halophenomenon effectively.②Low contrast is usually considered a serious problem. In this paper, a real-timeimage enhancement algorithm based on masking was proposed for low contrast images.Firstly, improve the global brightness through histogram stretching aiming at specificrange of pixels. Then local contrast is enhanced by applying masking. Related research work of this two algorithm is given in this paper. Also, detailedimplementation steps and algorithm flowcharts are presented. Many color images areprocessed using the proposed two methods and resulting effect is discussed. The resultsis compared with MSR and Meylan’s method. Parameters adjustment process and thebest choice of parameters are presented. All the resulting images and objectiveevaluation result data confirm that improved color image enhancement algorithm isachieved.