Adaptive Local Contrast Enhancement Based Tone Mapping For High Dynamic Range Images

The dynamic range is defined as a ratio from the gray value of the brightest part to the gray value of the dark part. The landscape in real world has a large dynamic range, even some can reach 10¹⁴, However, from the early film photography until the digital, their dynamic range is very limited, usually only 100:1 to 1000:1 magnitude, result to that it is difficult to accurately show the high dynamic range of reality scenes. With the development of digital image technology, particularly the technology of high dynamic range image, make accurate recording and reappearance of the scene in reality as possible. High dynamic range image ?HDRI? is a kind of which can represent the image type with the actual scene which gray area is relatively large, can be better to record and express scene in bright and dark areas in the optical properties. Now, majority of the image output device applies to low dynamic range, if we directly to the high dynamic range image on the display device to display, the show is different. So we need find a way of mapping to transform, to compress the dynamic range so that it can better show in the display device, this is the tone mapping study of the problem.In this paper, firstly, we can understand the background knowledge about the high dynamic range image through to interpret human visual characteristics and the registration synthesis of high dynamic range image. Secondly,we have introduce some previous tone mapping algorithms, and some classical tone mapping algorithms are analyzed, including the global tone mapping and local tone mapping algorithms. Later, this paper combined with this kind of idea about "dark places become darker, the bright becomes brighter", we gradually evolved into a locally adaptive contrast enhancement method, and then applied the method to high dynamic range images to tone mapping. We have proposed a tone mapping algorithm which based on adaptive contrast enhancement for high dynamic range images.The main idea of the algorithm is that the first is converted image to brightness and used adaptive logarithmic mapping to compress the dynamic range of the image, then get the compressed image to contrast stretching.In order to avoid the noise produced in the stretching process, we use fast bilateral filtering processing to preserving edge and reducing the noise after the stretching process. Finally, the brightness of the image is converted into RGB image.Experiments show that the algorithm proposed in this paper can be very good for the high dynamic range images. The algorithm can keep good overall effect of light and shade, and can better reflect the local images details and local contrast. The algorithm can be showing a more realistic scene in the conventional display devices, and has a lower average time complexity ?the average time complexity of single pixel is O ?1??. The picture which size is 512*768, for example, on the Intel ?R? Pentium ?R? CPU G630 @ 2.70GHz 2.70GHz equipment required calculation time:0.312 seconds. Through this mapping algorithm compared with other algorithms in this paper, whether in the treatment effect and treatment time have good competition.In addition, the paper gives the application of the algorithm in the general video enhancement, but also has a good effect.