Research And Application Of Low Illumination Image Enhancement

The image acquired from low light scenes has some bad features with low brightness,low contrast and much noise due to dim light and under-exposure.Moreover,it is so significant to research on low illumination image enhancement.Some enhancement algorithms are proposed on the basis of the characteristics of low illumination image and analysis of existed low illumination algorithm.Those are what have been done in this paper:(1)The algorithm which is based on the Retinex theory is combined.Low illumination image processing model which is based on SSR algorithm is summarized.Besides,method of low illumination image enhancement which is based on image segmentation is proposed.In HSV color space,we use guided filter to estimate component V in order to acquire the light component.Then the bright part and dark part of the light component are divided by image segmentation.The bright part of reflected components is not processed,which remains its detail.Meanwhile the dark part of the reflected component is denoised by the guided filter.The enhanced image is more bright,and overcomes the problems of halo and noise.(2)The defects of the traditional enhancement algorithm are analyzed and summarized,so that the bright channel is used in our algorithm.The component V is processed separately in the HSV color space,and the light component and the reflection component are estimated by the bright channel.So as to suppress the defect of the high light,an adaptive logarithm correcting method for the light component is further proposed.Through the comparison of the effect test and the enhancement effect of different algorithms,it is proved that the algorithm can improve the global brightness,suppress the bright part in the image,overcome the halo effect,and the image detail is clear while the color is natural without distortion.There is a very good visual effects in the image.(3)Low illumination haze image is analyzed and restored in this paper.Firstly,the haze image is reversed to the low illumination image.Then,the reversed image is restored,by the preceding algorithm.Lastly,turn back the restored image.The low illumination haze image is decomposed into low illumination image and haze image.The final reconstructed image is acquired by low illumination image enhancement,white balance and dehaze.This algorithm has low complexity,and the enhanced image is clear,and the color is natural.(4)The algorithm is successfully transplanted to the iOS platform.The original algorithm is verified,simulated and optimized in Matlab.We use OpenCV framework,and make the algorithm have its engineering package in VS2013.We load OpenCV framework,and transplant,debug,simulate our algorithm in iOS development platform Xcode.Through the effect display in the iPhone,the processed image has appropriate brightness,and has a strong contrast,showing more details,with good visibility.