Knee Osteoarthritis (OA) is one of the most common body complaints worldwide. As the prevalence of knee OA increases with age, the overall burden of this disease is expected to increase with the aging of the population. To assist doctors with accurate and reproducible assessment of knee OA parameters, we proposed an automatic approach using distance-based Active Shape Models (ASM) to calculate the geometric parameters between tibia and femur from an anterior-posterior knee radiograph. The proposed method employs the distance from knee joint boundary as one of the parameters in the cost function to cope with the weakness of traditional ASM, which is easily trapped by outlier configurations when approaching target boundary points. The experimental results show our system can derive these parameters accurately and efficiently.