The laser and camera calibration problem has been the primary issue in the subject of fusion of 3D space and 2D image information. While the solution for the calibration is mathematically well defined as closed-form by least squares techniques, reliability of the solution can be degraded by uncertainties in measurements. To enhance the reliability of calibration results, we adopted the EM (Expectation- Maximization) algorithm as a noise removal process in the sensor system. The simulation and real experimental results show the effectiveness of our approaches.