Study On Fine Recognition Of Single Material Granule Based On Laser Triangulation Photogrammetry

According to the development direction of modern close-range photogrammetry technology,directing at the identification and sorting problems of high-mixed materials existed in various industries,this paper selected the most complicated automotive shredder residues,which were generated in automobile recycling industry,as example,studied the fine measurement and identification of single material particles with the application of modern close-range photogrammetry technology and sensor-aided identification technology.The paper focuses on several key technologies include the design of laser triangulation measurement system,the application of image processing analysis technology,circularity analysis and multi-data fitting analysis.Combining with the research topic,technologies described above was applied to the automotive shredder residues identification and sorting,which was detailed elaborated in this paper.Main works of this article were summarized as follows:(1)The theory of single material granule recognition was expounded,key impact factors i.e.the thickness,particle size,and acoustic emission efficiency were determined.The experimental process and structure of the recognition system were designed and the preparation of sample,the experimental equipment of acoustic signal was presented.(2)The basic principle and thickness measurement process of laser triangulation was introduced and system contained camera sensor,measurement,data transfer and process was designed.Line scan CCD camera was selected for imaging and then multiple line images acquired by camera were combined and transferred into grayscale images.(3)Appropriate image processing algorithms were applied to the process of acquired grayscale image.The image was denoised using bilateral filtering,and binarized with local adaptive threshold segmentation,then the holes in targets were filled and the edge was smoothed using mathematical morphology operations.The connected domains were detected and labeled to match the data,finally,‘Canny' operator was used for edge detection.(4)Circularity measurement index was brought in to analyze and measure the target scraps in grayscale image.The advantages and disadvantages of several classical index were compared and analyzed and Roundness Factor was chosen as the most optimum index,then equivalent particle size of each scrap was determined.(5)RRSB function was used to analyze particle size/mass distribution of plastic scraps and the homogeneity of the scraps' particle sizes was verified.All the data acquired were fit and analyze to determine the characteristic parameters of each material,then the confidence level analysis was brought in to determine the theoretical recognition rate of the same matrix plastic,and compared the results of traditional sieving method.The results showing that by using the novel methods,recognition rate improved significantly.For mixtures of distinct metric like ABS and PP the theoretical recognition efficiency could achieve 100%;for ABS and ABS/PC mixtures,the theoretical recognition efficiency could approach to 90%,for PP and PP/EPDM mixtures,the theoretical sorting efficiency can approach to 70%,bringing about 15% and 20% improvement relative to traditional sieving method,respectively.