Research On Rapid Dimension Measurement Based On Image Processing

Tubular propellant grains are usually manufactured with screw extrusion technology. The consistency of their inner and outer diameter geometry dimensions is directly related to the performance of weapon launching system. Traditional measuring method uses micrometer to measure the diameters by manual. For extruded in production, the transverse plane of grain becomes the irregular ring. The manual measurement error is big, also time-consuming and labor-consuming.According to the production line structure, dimension measurement technology of tubular propellant grains'inner and outer diameters, based on multi-moving objects recognition, is proposed. Measuring project is designed and measuring system platform is built up. The experiment to samples proves that the technology reaches the requirement of the production line, and obtains higher precision.Five parts are mainly discussed: image pre-processing, image segmentation, multi-moving object recognition, sup-pixel dimension measurement, system calibration. In image pre-processing part, in view of big gray changes of the tubular propellant grain, compound filter, composed of fast median filter and mean filter, is presented, which preserves edge details while gaining fine smoothing effect. The weighted self-adapting threshold segmentation algorithm, based on single object'gray mean and variant, is introduced to solve the problem that the gradation change is so big within different tubular propellant grains that the segmentation is difficult. Three objects recognition algorithm, geometry parameter match, spatial moment match and Fourier Descriptors match, are studied and compared. For best matching result, improved 2-D polar coordinate-based Fourier Descriptors is adopted to realize the grains'recognition. In view of Fourier Descriptors'large calculation and low matching speed, a simple moving objects centroid forecasting means is used to forecast objects'location effectively. Fourier Descriptor matching is only adopted while the grain is going to leave the field of view. Sub-pixel edge location technology is a key component for precision enhancement. Spatial moment sub-pixel edge detection is adopted to realize sub-pixel measurement for tubular propellant grains, the precision is enhanced largely. Adaptive to the rolling of grains, a blocked calibration method is adopted, in which the calibration parameter changes with objects'different positions. The calibration process is simplified.Finally, dimension measurement experiment is carried out with the system platform for average inner and outer diameters of tubular propellant grains. The results indicate that the presented method is feasible and effective, the measuring consistency is fine and the precision is high. It is proved that the measurement error is lower than 0.02mm.