Binarization of images with variable lighting using adaptive windows

In this article, we present two algorithms for robust binarization of images with variable lighting; we call them Image Binarization with Adaptive Averages and Image Binarization with Adaptive Windows. Both algorithms have a decision criterion based on the local properties of the image to be binarized. This criterion calculates the size of a square window in each pixel of the image on which the characteristics allow us to discriminate to which class each pixel of the image belongs. To calculate the window’s size, a function is optimized that limits the window’s size considering the edges of the light and dark regions. The function is optimized using binary search. In both algorithms, integral images were used to have algorithms with execution times of tenths of a second.