In recent years, as society continues to face an aging population, there is a shortage of caregivers, leading to an increase in the demand for elderly care. Accurate assessment of the health status of the elderly and the provision of appropriate care requires timely recognition and analysis of human activities. To address this challenge, this paper proposes a Continuous Human Activity Recognition System that generates a 3D human skeleton model and utilizes joint angles to perform daily life activity recognition and infer similarities in movements across various body parts. The proposed system generates a 3D human skeleton model using depth information obtained from multiple RGBD cameras and extracts human joint angles based on this model. Furthermore, it utilizes time-series joint angle data of human activity to continuously recognize actions and estimate the similarity of movements in various body parts. To validate the effectiveness of the proposed approach, detailed verification experiments were conducted using real-world data.