Application And Research Of Image Processing And Recognition Algorithm In The Bullet Appearance Inspection

Bullet dimensions and defects in the traditional bullet factory is still detected by theinspector manual measurements and visual inspection, because such tests must take theinspector a lot of time watching the object to be tested, along with their own qualities,skills, experience and some objective constraints will appear devices undetected, theuneven quality and can’t accurately identify the special appearance of the imagedefects. With the rapid development of technology, factories intelligent degreerequirements are getting higher and higher, especially in the increasingly highmeasurement accuracy, efficiency and real-time requirements bullets industry is moreprominent.The traditional artificial detection means have not met the needs of theindustry. Therefore, how to solve these problems has become one of the hotspots.The rapid development of image processing and recognition technology is the keymethod to solve this problem. Based on the background of basic digital image processingand recognition algorithm, this paper analyses the characteristics of the bullet appearance,combined with the complexity of the detection, image processing and recognitiontechnology is applied to the outline of the bullet appearance of the image, the ly_size andthe surface defect detection.Firstly, introduce the related theoretical knowledge of image processing andrecognition and summary its applications. Then, put an image forward processingtechnology of bullet appearance, including image noise and filtering method, surfacereflection phenomenon and its processing and image enhancement technology. Comprisethe advantages and disadvantages of basic edge detection operator on the basis of theoriginal image filtering and enhancement processing. At last, propose one kind newvariable dynamic jd_yuzhiold contour extraction method based on the ant colony algorithm.Accurate measure the bullet’s length, width and the bottom circle diameter and otherkey parts of distance using the Harris corner detection and Hough straight line fittingmethod, and then make the error analysis. Finally, use the finite difference algorithm toextract the characteristic parameters of the defect region. For different defects regionaltectonic segmentation method, respectively connected region labeling method and areafitting method are selected to solve the area. According to the morphological knowledge,the bullet scratches can be classificated, measured and analyzed. Calculate the profile andly_size of point scratch and the length of line scratch. Finally compare the result with the standard values and then judge whether the product is qualified or not.Lastly, study the algorithm in this paper for simulation of MATLAB or programverification of VC++6.0and obtain the expected results. The experimental results showthat image processing and recognition algorithm in the Bullet AppearanceInspection can effectively overcome the defects of existing artificial visual method andhas higher superiority and applicability. More important is the stability and the accuracyhave been improved.