Community detection is a fundamental task in the study of complex networks, enabling the partitioning of nodes into distinct clusters to enhance network organization and analysis. However, many existing methods suffer from a common limitation: they require prior knowledge of the number of communities or a predefined community structure, hindering their applicability to diverse network scenarios. To overcome this challenge, we introduce a novel approach termed Leader-Similarity Community Detection (LSCD) algorithm for community detection in complex networks, leveraging leader nodes and considering the entire network structure. Our experimental evaluations on real-world networks demonstrate that LSCD outperforms state-of-the-art methods in terms of accuracy and robustness.