A High Speed Camera With Auto Adjustable ROI For Product’s Outline Dimension Measurement

In the industrial production, the three-dimension contour information of the products hasa very direct impact on the technical performance evaluation. Currently most domesticfactories along the coast still manually detect products’ contour to decide if they meetindustry standards. This method is costly, low efficient, and easy to misjudge the qualifiedproducts or miss the unqualified ones, thus seriously affects the factories’ efficiency andcredibility. Therefore, the majority of the factories are seeking transformation, expect tointroduce advanced technology, to realize the full automation of the factories. In this paper,we design a specific high-speed camera system with auto adjustable ROI for machinearbor’s outline dimension measurement. The entire system includes an illumination part, acamera part, a mechanical structure part, a signal processing part based on FPGA and PCsoftware processing part. The system will help factories to realize automatic arbormeasurement, and improve their efficiency and reduce their cost.The illumination part is a10×80LED pulse illumination array for a200mm×100mmdetection area. The challenges for this part are the high power and FPGA pulse signaldriving issues. The central component of the camera part is a2048×1088high-resolutionCMOS. The CMOS uses a ROI(Region of Interest readout) function, which automaticallyadjusts the CMOS exposure area as an arbor rolls on a tilted conveyor belt. With thisfunction, the CMOS frame rate can reaches2873fps and above. The mechanical part wasdesigned to service the whole system. An arbor rolls along a mechanical track. Thedetection and classification of an arbor is finished when it roll off the track. The detectionspeed is one arbor per second with accuracy0.1mm. When the arbor is in the systemillumination area, it blocks LED light, and forms a sharp shadow on the CMOS imaging. Inthis paper, we use FPGA for real-time signal processing. The FPGA calculates theminimum diameter of each cross-section of an arbor, restores its three-dimensional contourin real time using the CMOS image, and then determines if it is qualified. The PC softwareprocessing part shows the test results and image information, transfers control signals, andcompletes the friendly man-machine interactive.