HDR Fusion Algorithm Based on PCNN and over Exposure Correction

For traditional HDR fusion algorithm based on mapping functions, the artificial determination of the over exposure region often ignores the spatial distribution and the relation of pixels of the image brightness. In this paper, a stimulated HDR fusion algorithm based on pulse coupled neural network is proposed. For the proposed algorithm, first obtain the external stimuli according to the maximum image brightness, calculate whether the pixel “fires” or not and determine the over exposure region. Then alter the value of external stimuli to perform iteration and construct judgment matrix to make sure that every pixel in the image “fires”. The image irradiance is linear with the illuminance and exposure time. Obtain the brightness mapping function using linear fitting, and synthetize HDR image with over exposure correction algorithm to eliminate artifacts. The algorithm mainly has the following advantages: 1. It fully considers the spatial distribution of the image brightness. 2. For the imaging of different scenes, it self-adapts to separate exposure regions. 3. It performs mapping according to the imaging principles and obtains the mapping function without the exposure time. 4. It reduces the quantitative error of synthetizing image, and the artifacts in the transition region. The algorithm proposed in this paper is experimented under multiple scenes, and the image entropy and spatial frequency are improved greatly.