For achieving non-contact high-precision and rapid online measurement of the surface contour and key geometric dimensions of moving workpieces on a roll forming steel production line, the paper presents a roll-formed part online measurement system based on multi-view vision and structured light. Initially, the system achieves high-precision global calibration using a planar target and multi-objective calibration principles. And then the spatial pose relationship of multi-head measurement coordinate system space was calculated. Subsequently, second-order algorithms were employed for precise image projection correction to accomplish 3D reconstruction. Finally, the center of light stripes at sub-pixel level was extracted. Firstly, a pruning algorithm was utilized to eliminate random errors along the stripe center line, and then obtain precise point cloud data of the structured light, next reconstruct the workpiece's cross-sectional features and measurement of its bending angles and outer contour dimensions, and finally complete online monitoring of workpiece dimensions. Experimental results demonstrate that the measurement accuracy of this online measurement system is 0.05 mm, which indicates its high precision and accuracy. This system has promising applications in the roller-forming industry and significantly improving workpiece detection efficiency.