Despite the widespread use and benefits of knowledge graphs (KGs) in diverse fields such as question-answering, recommendation systems, and natural language processing, they indeed possess certain limitations. Two of the most critical challenges associated with KGs include the presence of noise and incomplete data. The presence of noise and missing data can substantially compromise the effectiveness of applications that utilize these KGs, potentially leading to incorrect conclusions or decisions. These challenges highlight the need for robust methods to enhance the quality of KGs, particularly in terms of their completeness and trustworthiness. Thus, it is important to assess the quality of knowledge graphs in terms of trustworthiness and completeness. Additionally, addressing these issues through automated methods is a vital step toward improving the performance and utility of KGs in a wide range of applications. Firstly, we propose a new approach to automatically evaluate and assess existing KGs in terms of completeness and trustworthiness by assigning specific scores. Secondly, we present, TrE-ND, an ensemble learning method for noise detection in KGs, which outperforms state-of-the-art methods designed for noise detection in KGs. Consequently, our proposed method offers a practical and efficient resolution to the persistent noise issue in knowledge graphs, which improves the quality of knowledge graphs in terms of trustworthiness. Finally, we present a dynamic selection mechanism to form an efficient ensemble for noise detection in KGs.