The article introduces a new method for improving image segmentation by addressing the issue of image noise and the limitations of traditional threshold selection techniques. The researchers developed an algorithm that corrects image noise by adjusting the three-dimensional histogram and selects thresholds based on gray entropy decomposition. This approach reduces noise interference and computation complexity, resulting in better segmentation quality and faster processing time compared to existing methods. The algorithm outperforms other techniques in terms of noise resistance, visual quality, and efficiency.