A HDR Exposure Compensation Algorithm Using Piecewise Tone Mapping Functions

This paper presents a HDR exposure compensation algorithm for less pleasant images, acquired with poor or excess illuminations and backlightings. General exposure corrections are classified into local and global approaches. Retinex is the most typical method of local ones, that a lot of surround pixels are utilized and the correction is implemented per pixel. Thus the results of the locals are pretty good but much computation time is required. In contrast, a tone mapping, known as gamma correction, is the most representative method of global ones. They are relatively simple and fast because of using a global mapping function, but locally detailed features are not considered. Therefore, a piecewise tone mapping algorithm using the statistical distribution of its input is proposed to obtain HDR images efficiently. The two domains of the proposed piecewise mapping functions are divided by the global mean, and the gamma orders of the functions are calculated due to the class means of dark and bright side, respectively. As the result, since the proposed algorithm uses only 3 parameters, such as the global mean and two class means, it is simple but can obtain a good output about the asymmetric distribution images even with severe exposure differences.