Image Enhancement Algorithm Based On Multi-Exposed Images

As people's living standard is improving, leisure life becomes more and more colorful, making the camera more and more indispensable. In real life, the human eye has a wide observing range. However, for ordinary digital cameras affected by the hardware devices, sensitivity range is limited. when the span of the scene brightness range is wide, serious loss of part of the details in high-brightness part of shadows results in digital cameras'disability of obtaining all the light intensity information for a single photo, even if adjusting the camera's exposure time.To solve this problem, this thesis proposes a multi-exposure image enhancement algorithm which utilizes a group of pictures with different exposure time for the same scene. First, pyramid decomposition is carried out for each of the pictures, afterwards, the decomposed image is partitioned into blocks, of which useful ones containing much information are selected to compose a picture. Then the fusion algorithm is used to eliminate the block edge discontinuities of the result picture. The final output image contains a large dynamic range of brightness, achieving the object of enhancing images. The core of this algorithm is pixel block selection and integration. In order to reduce calculated quantity to improve the computing speed, this thesis uses pyramid decomposition of the image first to reduces the image size. For the image block selection, the camera's response curve is referenced to measure the amount of information of the image blocks. And for the block fusion of the image, both the amount of information and continuity of pixel blocks are considered to obtain better fusion effect.This thesis shoots multi-exposed pictures as the data sources, compares the results gained from several existing image enhancement methods and the method proposed in this thesis. What's more, both subjective and objective methods are used to evaluate the enhancement methods. The feasibility of the proposed method is high, additionally, it is easy to calculate and improves the information capacity of a single image by enhancing the details of the image.