The efficient management of traffic flow on highways is crucial for ensuring safety, reducing congestion, and optimizing transportation infrastructure. However, real-world scenarios often involve missing or incomplete traffic data, posing challenges for decision-making in Intelligent Transportation Systems (ITS). In this paper, we present a novel methodology, the GAN-Transformer Imputer (GTI), for spatial-temporal traffic imputation. Our approach integrates Generative Adversarial Networks (GANs) and Transformers to address the limitations of existing methods. The GAN component generates realistic traffic data, while the Transformer captures long-range dependencies and temporal patterns. Through a series of experiments on synthetic and real-world datasets, we demonstrate the superior performance of GTI compared to state-of-the-art methods. GTI consistently outperforms traditional statistical methods, machine learning approaches, and deep learning methodologies in terms of Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and F1 Score. The key contributions of this work lie in the seamless integration of GANs and Transformers, providing a robust solution for spatial-temporal traffic imputation. Our methodology, with its adversarial training and uncertainty quantification, holds promise for enhancing decision-making processes in ITS, contributing to the advancement of transportation systems and urban planning.