Image Binarization with Hybrid Adaptive Thresholds

Abstract

AbstractThe process of image binarization divides pixel values into two groups, with the back-and-forth represented by white and black, respectively. A novel image binarization approach is presented in this research, “Image Binarization with Hybrid Adaptive Thresholds (IBHAT)” with two adaptive thresholds being proposed. It associates two threshold values which are based on global and local mean for each pixel that is to be transformed to either in background or foreground. One of the threshold values has to be selected for the concerned pixel. Hence the term hybrid adaptive is used. The local contrast of pixels inside a block of 3\(\times \)3 is used to make the decision. This technique can switch to three different modes such as local, global, or hybrid adaptive thresholding technique simply by using a specific control parameter value. Global adaptation technique gives a result similar to global techniques, local technique gives a result similar to other local techniques, while the hybrid technique gives a result which is not similar to the previous two but is a very effective one. So it is convenient to apply in degraded document image binarization. This technique is compared with other global as well as local techniques at different conditions of the controlling factor based on the types of the image applied. Since it uses only 3\(\times \)3 block size, its computational time is similar to the global one. In comparison to previous relevant strategies, the suggested methodology takes less time to compute and produces effective results, according to the results of the experiment.KeywordsAdaptiveBinarizationGlobalHybridLocalThresholding