A new method for improving image segmentation has been developed. This method, called MAOTSU_2D, enhances the traditional 2D Otsu's method by using a combination of median and average filters to smooth the image. By creating a 2D histogram and performing two one-dimensional searches, the optimal threshold vector is determined. Additionally, a post-processing step is introduced to handle noise and edges in the image. The MAOTSU_2D method shows significant improvements in dealing with Salt&Pepper and Gaussian noise compared to the traditional 2D Otsu's method. Experimental results on both synthetic and real images confirm the effectiveness of this new approach.