Research On ROI Localization And Defect Detection Of Plate Type Heat Exchanger Image

Plate type heat exchanger is a kind of heat exchange device for heat exchange,which is made up of a series of metal plates with a certain corrugated shape.The quality of the plate determines the effect of heat transfer,and the cracked plate will cause safety accidents,but also the cracked plate may even cause safety accidents.Whether the quality of a plate is qualified or not will directly affect whether a heat exchanger can enter the market.In the process of production and pressing,the surface defects of corrugated plates are caused by uneven gravity stamping.According to the different degrees of defect cracking,the defects are divided into crack defects and "double eyelid" defects.At present,the defect detection of plate type heat exchanger plates in industrial production mainly depends on manual sampling and penetrant flaw detection methods,but these methods have low detection efficiency and are affected by the subjective consciousness of inspectors,so they cannot meet the requirements of fast on-line detection.With the continuous improvement of the application of machine vision,Shenyang University of Technology has developed an on-line visual inspection device for corrugated plate defects,Complete the acquisition of the plate image and the defect detection of the fixed threshold for some areas of the plate image,However,the fixed threshold method can not meet the inspection requirements of hundreds of plate models.Aiming at the defects in the production process of plate type heat exchangers and the understanding of the defect occurrence area by prior knowledge,this paper studies the plate image region of interest location and defect inspection methods of plate type heat exchangers to meet the needs of existing productivity.In this paper,firstly,the image of the plate collected in the production site is preprocessed,morphological operation and geometric feature screening are used to reduce non-plate areas and minimize the influence of interference areas on detection.According to the characteristics of ripple areas and diversion areas,ripple areas and diversion areas are located and clipped,and the plate image is divided into three large blocks to shorten the running time.Then,the region of interest area is divided according to the cause of defects,and locating the region of interest area can greatly reduce the workload of computers and increase the detection accuracy.Optimize Canny classical edge detection algorithm and improve it into an adaptive edge detection method that automatically adjusts parameters according to different plate shape characteristics.This method has remarkable effect in ripple area positioning;For the diversion zone with complex structure,by merging the maximum and minimum points in the edge,only a pair of maximum and minimum points are retained in the edge,the relative amplitude of each column in the merged image is calculated,and the threshold segmentation point is found by statistical analysis.This method is only based on the placement point of the edge minimum,so that the location is no longer limited by parameters.Finally,the defects are detected by gaussian line in the localized region of interest area,and the image enhancement improves the influence of gaussian smoothing on the contrast of defects.For fine defects,the local maximum method is used to describe the defects,and the false detection interference is suppressed to reduce the false detection rate of defects.Based on machine vision,this research uses image processing method to provide a new solution for plate inspection,which has the advantages of accuracy and efficiency in inspection.