The emergence of the Internet of Vehicles (IoV) signifies a transformative paradigm for intelligent and interconnected transportation systems. However, the effective sharing of data within the IoV ecosystem presents significant challenges including security, privacy, and efficiency. To address these challenges, this paper introduces the IoV-SFL, a novel framework that seamlessly integrates Federated Learning (FL) and blockchain technology to provide a comprehensive solution. By employing federated learning, IoV-SFL framework ensures that data remains on local devices, preserving user privacy, while consortium blockchain enhances security and transparency. Furthermore, the IoV-SFL framework also leverages the capabilities of Homomorphic Encryption (HE) and Dynamic Scaling Factor to enable privacy-preserving computations, optimize resource allocation, and enhance efficiency. The incorporation of Gated Recurrent Units (GRU) within the Convolutional Neural Networks (CNN) model empowers the IoV-SFL framework to extract meaningful insights from heterogeneous data streams. Rigorous experiments have validated the superior performance of the IoV-SFL framework compared to conventional approaches in terms of security, communication efficiency, model accuracy, and convergence speed. Additionally, a comprehensive security and privacy analysis highlights IoV-SFL's robustness against diverse threats and vulnerabilities. The IoV-SFL framework represents a groundbreaking advancement, offering a secure and efficient means of data sharing in the Internet of Vehicles (IoV), paving the way for innovative vehicular applications and safer transportation systems.