Abstract: As camera calibration results can be influenced by external factors easily, to improve the accuracy of calibration results, the pin-hole camera imaging model was utilized, and the adaptive corner detection method was adopted to pick up the feature points in calibration images, then the mean value of re-projection error in the coordinates was introduced as the performance index, a series of experiments were designed correspondingly for the three external factors including the number of image, the luminance of light and the size of square in calibration board. The experiments results indicate that choosing brighter light source can improve the calibration accuracy above 38%; given 156 points feature, only 18~22 calibration pictures are needed, and also, smaller square size can increase the results- accuracy about 50%. All mentioned above show the importance of light source, number of pictures and square size in camera calibration accuracy-improving.