The processing process of OD travel matrix based on traditional traffic surveys requires a lot of manpower, material resources, funds, and time, and cannot be carried out frequently. Cell phone data, as an important supplement to existing transportation data collection methods, provides excellent technical support for extracting OD features of residents' spatiotemporal travel. At present, the stay point judgment for mobile positioning technology mainly focuses on the travel status judgment of cell phone data time series, ignoring the characteristics of user travel activities and travel purposes, such as general user travel activities such as commuting, life, entertainment, and travel. In response to the difficulty in identifying different travel activities, a low-cost and highly accurate method is proposed for determining stay points based on job residence correspondence and spatiotemporal kernel clustering method. The systematic stay points extraction approach, combining with work and residence and POI data analysis, can improve accuracy and rationality of cell phone users’ travel activity characteristics by eliminating signal roaming points and dwell points with central clustering characteristics in continuous short distances through spatiotemporal kernel clustering analysis. The systematic stay points extraction approach combined with spatial and temporal characteristics is implemented in Guangzhou city, results show that travel activity OD pairs given by the systematic stay point extraction approach can clearly presents different travel activity characteristics of users.