A new method for improving image segmentation accuracy and universality was developed using a modified 2-D Otsu thresholding approach. By adjusting the probability distribution and calculating object and background areas more precisely, the method achieved better results with less memory and time compared to existing methods.