The curves on well-logging parameter graph are very important to reservoir description. But how to keep the parameter graphs permanently and use the data implied in those curves efficiently are still open problems in petroleum industry. In this paper, we contribute an approach to digitizing well log curves and storing the digitized information into Oracle database or data file server with multi-agent perspective. We employed Gaia methodology and open agent architecture to analyze and design the system. According to the characteristics of the well-logging parameter graphs, we implement the SCTR (Scanning, Compressing, Tracing, and Rectifying) algorithms with four agents to digitize well log curves. Two data management agents are developed for operating database and data files in uniform agent communication language. The experimental results show that this approach is effective.