AnAlgorithm forHigh-precision Edge Location ofCannonBarrel ImageBasedon HumanVisual Characteristics

A novelalgorithm forhigh-precision edge location basedon humanvisual properties isproposed considering characteristics ofcannonbarrel images, which improves theobservation effect ofspying boreimages and diminishes thetarget imageambiguity caused bybackground andnoise. Thevariable grayscale areamethodisapplied according totheimagefeature diversity between thetarget andbackground/noise. Theaccuracy ofedgedetection and location ishigher than0.01pixel whenbeingapplied for engineering. The imperfect imagesare correspondingly enhanced, andthetarget isthusshownclearly, whichis convenient forobservation by human eyesand actual measurement. Thealgorithm ismoregeneral-purpose and adaptive todifferent barrel imageprocessing, anditcan significantly inhibit noise forits glossy andenhanced effect. I.INTRODUCTION INVESTIGATIONSandapplications ofnon-contact measurement techniques forthecannon static parameters measurement areintheir beginning stage intheworld. The purpose ofatesting system forcannonstatic parameters measurement istofindtheimperfection ofthecannon barrel images collected primarily bythesubmicron and high-accuracy CCD camera, suchasablated parts, crack, breakout, impression, etc., andtoanalyze andcompute its allparameters bylighting control, imagepreprocessing, image picketage andsystem control. Intheprocess ofcannonbarrel inspection, itisvery important toacquire theaccurate information ofthecannon barrel images. Because theinfluence ofillumination, sensor resolution, thespecial construction ofcannon bodypipe can maketheimaging quality poor, ithastofind aneffective algorithm foredgedetection andlocation tosatisfy the characteristics ofthecannon barrel images.