This study employs bibliometric methods to analyze the research landscape of affective computing in human-computer interaction (HCI). Using data from the Web of Science Core Collection and visualization tools such as VOSviewer and CiteSpace, it elucidates the current state of research, theoretical foundations, key issues, and trends. Findings indicate a three-stage evolution with a notable increase in publications since 2019. Research has expanded from computer science into interdisciplinary areas including psychology, sociology, biomedicine, and neuroscience. Key contributions come from countries like China, the United States, the United Kingdom, and Germany, with leading institutions such as MIT and the University of Amsterdam. The theoretical framework encompasses four interrelated elements: emotion understanding, recognition, modeling, and databases. Current research trends focus on emotional interaction, deep learning, physiological signals, and multimodal fusion. Future studies are expected to enhance the intelligence and naturalness of HCI, improving user satisfaction and interaction quality.