Evenly brightening using kurtosis Gaussian pattern to simplify image binarization

Abstract

Abstract A problem of separating object from background using thresholding in image segmentation is uneven image brightness. Some binary image objects vanish because the pixel intensity value on the edge side is greater than the pixel’s background intensity. The others at the edge of image are covered by background pixel. The background pixel at the edge is bigger than the pixel value of the object in the inside of image. A light source type affects the spread of brightness adding thresholding difficulty. The object’s light distribution from the exposed light source follows the Gaussian distribution. A kurtosis value depends on the light source type. Following the kurtosis of standard deviation, the method for creating brightness evenly is designed. The Gaussian pattern is formed from the distribution of Gaussian values in the cells of the Gaussian matrix. The pixel values in the corresponding images will be added to the matrix cells. The pixel value will increase according to the Gaussian matrix pattern. This process results in the highest pixel value not getting added value, while the lowest pixel intensity value will get the greatest value addition.