Research Of Image Processing For Recognizing Potato Strips' Quality

Potato strips,with high nutritions,can bring amounts of money,which become more and more welcome in the world.However,the techniques for processing potato strips,especcially on grading strip quality,still has a long way to a completely mechnized producing line,except the machines from abroad. By the technique of image processing to grade strips and get rid of manual work,this thesis will prvide some theoretical and experimental referrence for machined mass producing.The process of grading potato strips is the one of recognizing the outside quality of potato strips by their characters in shape,size and color .Firstly,gain their original images in BMP image format,which comes true by a programming module to turn the five ordinary image-file formats(PCX,GIF,TIFF,JPEG and TGA) into BMP and also make this sofeware more useful.Secondly,separate strips from their backgrounds by the automated most-variance method ;filter these noises and small impurities with the knowledge of math configuration;smooth the strip images with median filter wave and through a seed-filling mode,fill colors in bright spots and holes.Then,based on filamented and cutting-branches methods, get five shape charaters and two color ones:one length,three widths,one curvature,average and standard squared subtract in hue,Finally,dependent on the knowledge of mode recognition,fix seven thresholds about these characters to grading these potato strips.After testing 216 pieces of strips.this sofeware has a precise-recognizing of 83.8%,and a justice miscarriage of 16.2%.This research develops a sofeware system applying in recognizing potato strips,which has powerful functions and a friendly interface.