Research Of Image Enhancement Based On Nighttime Photographs

Recently, with the development of mobile networks, taking photographs has become an important way of socializing online. One special condition is taking photographs at night, which occurs more frequently. And the quality of the nighttime photographs on mobile devices needs to be greatly improved subject to the user’s interest and attention.Limited by the size of the mobile terminal hardware and the cost of design, many cameras on the market still have two major drawbacks in terms of the nighttime photograph:the noise and the blur. In addition to upgrading the hardware level, the thought of enhancing the quality of night shooting images by processing algorithms has been developed rapidly. A large number of postprocessing algorithms have been applied on the operating system of the mobile intelligent terminal to achieve the improvement of the photographs quality.This paper mainly studies two aspects of image processing algorithms as follows so as to improve the overall visual quality of image taken at night:Firstly, we study the common image enhancement algorithms, and propose an improved image texture enhancement algorithm based on the theory of image decomposition and fusion. At the same time, we propose a new concept of multi-scale which is no longer the size and the resolution of the image in the traditional sense but the degree of image edge-preserving filter on the basis of the original algorithm. Meanwhile, the improved algorithm applied a filter which has better effect. The test results of the algorithm according to the comparison as well as observation are shown in the figure, which proves that the texture enhancement of the algorithm presented in this paper outperforms than several common algorithms, furthermore, the ring halo phenomenon—a common problem for other image fusion algorithms in the edge does not appear in our algorithm. This algorithm can be applied to multiple images fusion for image enhancement and help to improve the overall effect of the fusion algorithm.Secondly, we propose an improved method for nighttime photographed image enhancement which uses multi-frame fusion of a camera flash. Flash images have the abundant environmental information and details. However, the overall illuminationchanges. No-flash images are within noise and fuzzy details while keeping the original environment lighting condition. Therefore, this algorithm attempts to combine the texture detail of the flash images with the illumination of no-flash ones together. Based on image decomposition algorithms and fusion technologies, this algorithm applies a multi-frame and multi-scale image decomposition algorithm to extract the rich texture information in the multiple flash images, and adopts multi-scale smoothing algorithm for no-flash photography image de-noising, and fuses the rich texture with the de-noising results. Compared to other low luminance image enhancement algorithm, this algorithm makes full use of the advantages of flash and no-flash images and removes unwanted noise while getting a lot of texture details. In addition, in order to enhance the performance of the overall algorithm, this paper proposes an improved contrast enhancement algorithm, which is suitable for the low luminance image. This algorithm could maintain the overall illumination of the image while enhancing the local contrast. It can not only improve the visual effect, but also help improve the quality of the image registration algorithm.Finally, this paper gives the implementation process, experimental results and the image quality evaluation method used in this algorithm. It can be shown by analyzing the experimental results that the algorithm presented in this paper produces visible effects, removes the noise obviously, obtains the clear texture, and enhances the overall visual experience of the original image. This algorithm can be widely used at nighttime photographs based on mobile terminal.