遗传算法是当前许多科学实验领域广泛应用的一种非线性并行算法.本文研究了遗传算法在数字图像的灰度图二值化中的应用 ,提出了一种新的灰度图二值化方法.该方法通过对每个子群体的优化计算和动态改进的适应度函数 ,确定新的分割阈值.实验验证该方法对于噪声干扰的一般质量图像有着良好的效果.-Genetic algorithm is currently widely used in many scientific experiments in the field of a nonlinear parallel algorithm. In this paper, genetic algorithm in the digital gray level image binarization of the application, proposed a new binarization of grayscale method. The method by optimizing each sub-group improved calculation and dynamic fitness function, to identify new segmentation threshold. experimental verification of the method for the general quality of the image noise has a good effect.