Based on vision measurement technology, an online system for steel strip surface defects detection was developed. The sensor system consisted of CMOS line scan cameras and red LED line light source. The system could be used to achieve automatic detection of steel strip surface defects. The method to adjust the exposure time of line scan camera was presented by which the exposure time of the line scan camera could be adjusted according to the gray values of the captured images. The image processing algorithm was studied to complete the contour extraction of defects. Experimental results show that the system can be used to realize online detection for steel strip surface defects promptly and accurately. In this way, the steel strip quality can be improved.