The efficient segmentation of foreground text information from the background in degraded color document images is a critical challenge in the preservation of ancient manuscripts. The imperfect preservation of ancient manuscripts has led to various types of degradation over time, such as staining, yellowing, and ink seepage, significantly affecting image binarization results. This work proposes a three-stage method using generative adversarial networks (GANs) for the degraded color document images binarization. Stage-1 involves applying discrete wavelet transform (DWT) and retaining the low-low (LL) subband images for image enhancement. In Stage-2, the original input image is split into red, green, and blue (RGB) three single-channel images and one grayscale image, and each image is trained with independent adversarial networks to extract color foreground information. In Stage-3, the output image from Stage-2 and the resized input image are used to train independent adversarial networks for document binarization, enabling the integration of global and local features. The experimental results demonstrate that our proposed method outperforms other traditional and state-of-the-art (SOTA) methods on the Document Image Binarization Contest (DIBCO) datasets. We have released our implementation code at https://github.com/abcpp12383/ThreeStageBinarization.