Study About The Methods Of Image Mosaic And Cylinder Defect Detection For Online Product Inspection

With the development and put forward of the concept of intelligent manufacturing,manufacturing industry have made greater demands on online product inspection.The application of machine vision technology in online product inspection will be an inevitable direction.This paper start with defect detection and image mosaic method,researching the application of machine vision technology in online product inspection,the main content are:Analyze the existing problems in the scanning process of LED dies,and optimize it.The paper put forward a continuous shooting scanning mode based on image mosaic method,it can greatly improve the efficiency of scanning process.And analyze different combinations of different scanning mode from accuracy and efficiency,prove the effectiveness of the proposed method by experiments.Analyze the existing problems,which the defect Area is incomplete,in the defect detecting process.The paper using image mosaicing method for cylinder and defects stitching,put forward and compares two kinds of cylinder unwrapping method,and uses a way that stimulating multiple camera shooting,combines image unwrapping process and image mosaicing process together,prove the effectiveness of the proposed method by experiments.Analyze the cylinder detection item of motor commutator.First,the paper put forward an edge detection method that using sub-pixel threshold segmentation method replace traditional edge detection method,then using the sub-pixel point,which is after filtering,for line fitting.Then analyze the groove defection,put forward an automation threshold segmentation method which can segment more than one region,and combines the characteristics of the groove defection,it can detect the groove defect effectively.And then analyze the pit defect of cylinder,put forward a kind of stable region detection method based on local threshold segmentation,which can detect the full defect area.Finally,analyze the line features on cylinder,the line feature extraction method is based on differential geometry,and the paper improve it by differential singularities,hysteresis threshold connection and local contrast filtering method,experiment shows that it can extract different line defection effectively.