Image Specification Based on Multimodal Gaussian-Like Function

In order to solve problems such as histogram equalization’s uncontrollability, histogram specification’s subjectivity, and trial and error, a new image enhancement algorithm using histogram specification based on multimodal Gaussian-like normal distribution function is proposed in this chapter. The proposed algorithm calculates the average gray value and contrast of original image and uses these two parameters to obtain all means and standard deviations of multimodal Gaussian-like normal distribution function and then it specifies the histogram of the image processed with this function. In addition, by introducing the gains of Gaussian peaks, the wave shape of this function mentioned above can be adjusted. So with multimodal Gaussian-like normal function as histogram transformation model, the histogram of image processed is constrained into the specified shape. Finally, the experimental results show the superiority of the proposed algorithm compared with some existing algorithms such as traditional equalization, equalization based on bimodal Gaussian-like distribution, etc. It also has many advantages over other image enhancement methods such as low computational complexity, high efficiency, and no manual intervention. Moreover, it can be easily controlled. The algorithm can not only selectively enhance the image local contrast but also improve the average gray value; so the processed image looks more realistic, being rich in layers and details.