A new approach based on stereo vision technology is introduced to analyze sheet metal deformation. By measuring the deformed circle grids that are printed on the sheet surface before forming, the strain distribution of the workpiece is obtained. The measurement and analysis results can be used to verify numerical simulation results and guide production. To get good accuracy, some new techniques are employed: camera calibration based on genetic algorithm, feature abstraction based on self-adaptive technology, image matching based on structure feature and camera modeling pre-constrains, and parameter calculation based on curve and surface optimization. The experimental values show that the approach proposed is rational and practical, which can provide better measurement accuracy with less time than the conventional method. By measuring the deformed circle grids that are printed on the sheet surface before forming, the strain distribution of the workpiece is obtained. The measurement and analysis results can be used to verify numerical simulation results and guide production. To get good accuracy, some new techniques are employed: camera calibration based on genetic algorithm, feature abstraction based on self-adaptive technology, image matching based on structure feature and camera modeling pre-constrains, and parameter calculation based on curve and surface optimization. The experimental values ​​show that the approach proposed is rational and practical, which can provide better measurement accuracy with less time than the conventional methods.