Research On Purple Fringing Correction Method In Digital Imaging System

In recent years,imaging modules on the digital camear,mobile phone,monitors developed rapidly,shooting high resolution,high quality images has become the focus of digital photography,monitoring and other imaging-related industries.With higher resolution,image quality problems become more prominent,including noise removal,color correction,high dynamic range,etc..The purple fringing(PF),some object edges'color looks purple sometimes partial,will appear in some areas in the images taken by camera or imaging system under certain conditions.The PF belongs to color errors,correction of the PF is of great significance for improving the image quality and visual effect of the images.There are a variety of speculations for the reasons of the PF,including the lens chromatic aberration,demosaicing algorithms,over exposure of the CCD/CMOS devices,etc..In this paper,we firstly analyze the rationality of these interpretations,and then consider the color channel data of the PF regions in the images.Confirming the true cause of the PF is of great significance to the PF correction and imaging system.The existing methods to eliminate PF are almost image processing algorithms.There are a lot of researches on the PF correction by algorithms,which can be roughly divided into two parts,the detection of the PF and the PF correction,respectively.The potential PF needs to be marked in the detection step to serve for the implement of the correction method,which will correct of the PF's color errors and restore the color in the region,and ultimately enhance the visual effect of the image.There are three steps in the PF detection,including searching for the highlight area,extending the highlight region,and using the chromaticity diagram to mark the purple areas.The three steps were studied and improved respectively.Weighted information were added to judge highlight areas to enhance the robustness,the chromatic theory is exploited to extend the highlighted region to improve the accuracy,the modified chromaticity diagram is adopted to determine the purple regions.The traditional correction method can only be carried out on the small PF areas,or is easy to cause error correction.To address these problem,in this paper,we propose a new method based on the object color information to correct the PF,and rules to avoid the error correction are elaborated designed.At the same time,this paper studies the effect of the removal of remaining PF and the influence of image noise,and fast implementation of the method.Finally this paper puts forward two evaluation indexes to evaluate the proposed PF correction method,comparing with the existing methods,including pixel ratio moved out from the PF areas in the chromaticity diagram and the reduced degree of purple color of the PF areas.The effect of the PF eliminating algorithm can be reflected by the two indexes.The experiment results show that the proposed method has a great improvement on the indexes and visual effects.