As a new high-quality welding technology，Laser welding has been widely used in automobile, aviation, shipbuilding and other fields, and has great significance to the equipment manufacturing in China. Because of some time-varying factors in welding equipment and crafts in the laser welding process, there are some fluctuations in the quality of welding. So it is necessary to conduct online real-time weld quality inspection, which can not only guarantee the quality of welding, but also contribute to the implementation of welding automation. At present, real-time inspection of weld quality in laser technology has become a key part in advanced laser welding system abroad. But the technology is still in research stage at home. Therefore, the development of real-time weld quality inspection system used in laser welding is still of great importance in the academic theory and engineering applications. With reference to the advanced weld quality inspection system, the research about real-time inspection of weld quality is conducted to solve the scientific problem that exists in the practical application and to meet the demand of the practical application. In the first part of this thesis, firstly the weld quality inspection system based on the principle of structured light is introduced. Secondly, it focuses on the research about the image processing algorithm of the weld quality inspection system. Image processing algorithm is the core of the inspection system, its processing speed and accuracy influence the accuracy and real-time of inspection system directly. Image processing algorithm is designed based on structures light and ISO13919-1 B-class quality standards. Algorithm is divided into three steps: Firstly, the centerline is extracted accurately from Structured light patterns. Secondly feature points are determined from centerline. Finally, according to the geometry of the centerline and the extracted feature points, the parameters of weld defects are detected and calculated. The second part at first introduced the principle and method of structured light vision sensor calibration. The direct calibration method of vision sensor was studied specially for weld quality inspection applications. Furthermore the calibration method is designed. A linear function relationship is established between plane formed by laser light and image plane in tiny region so that any physical ordinate of the point in the image plane can be acquired through linear interpolation. Experimental results show that the accuracy of the direct calibration lives up to the accuracy requirements of quality inspection system, and is convenient to implement in engineering. In the third part of this paper, the software and hardware of weld quality inspection system are designed according to the demand of quality inspection of laser welding. Some key issues in the software implementation are studied, such as the method to achieve inspection systems by Windows, the priority of process and thread, the three-dimensional reconstruction and so on. Finally, the experiment is conducted to verify the accuracy and speed of the image processing algorithm in the inspection system, experimental results show that the image processing algorithm meets the system requirements. Finally, the research work is summarized，and the future work is prospected.