As access behaviors grow more dynamic and complex especially in cloud-based and distributed environments, traditional access control models are inadequate in terms of flexibility and risk perception. To improve real-time decision making in access control, trust mechanisms serving as key means for assessing subject credibility are increasingly integrated into access control frameworks. In recent years, numerous trust-based access control models have been proposed, yet a comprehensive review synthesizing and comparing these efforts is still lacking. This paper proposes a five-dimensional analytical framework, including trust modeling methods, authorization decision mechanisms, technology integration paths, environmental adaptability, and experimental validation to classify and systematically evaluate representative models. The paper systematically compares different trust modeling methods and authorization strategies, evaluates their implementation paths in blockchain, artificial intelligence, and other integrated technologies, and further summarizes common challenges in adaptability, interpretability, and evaluation approaches. In the field of trust-based access control, we intend to serve as a reference and guide for future researchers and practitioners.