Parameter estimation of image gamma transformation based on zero-value histogram bin locations.

Gamma transformation plays an important role in image processing and image display. However, images may look either bleached out or too dark if transformed with improper parameters. While the value of gamma transformation parameter with which an image was created is not included in the current image standards, it is hard to restore the original image vision without the knowledge of gamma transformation. In this paper, a gamma transformation parameter estimation method is proposed based on Zero-Value Histogram Bin (ZVHB) locations. The method firstly exploits the relationship between the number of ZVHBs and the parameter of gamma transformation to get an approximate value of the parameter and an interval to which the parameter belongs. Then in the interval, the parameter is searched from the approximate value by matching the ZVHB locations of the investigated image and the reference images transformed with two close parameters. The experimental results validate the effectiveness of the proposed method, which outperforms the state-of-the-art methods. Meanwhile, the improperly transformed images can be restored visually by applying inverse gamma transformation with the estimated parameters.