Cotton Yarn Blackboard Computer Image Recognition

Traditional grading of cotton yarn applies visual assessment of yarn onSerip1ane, which has certain weakness. In this paper, a new method of testingSeriplane based on compther image processing technology is proposed and deeplystudied.On the basis of careful market-investigation, the preset Seriplane method, it'sstandard and the feasibility of testing Seriplane through computer were discussedand anayzed. First, through comparative analysis, this paper pointed out weaknessof Serplane method. Then theoretical basis and physical essence of Seriplanemethod were discussed, from the point of visual optical principle. Secondly, thekey reason Why Seriplane method is superior to instrument was given, the processand mechanism of visual assessment was also revealed. Finally, a model andconditional parameters of intimating visual assessment through computer was put.Besides, three possible method for processing image of cotton yarn Seriplanethrough computer were given, which are morphological, direct dividing areamethod and co-occurrence matrics.In view of the feature of Seriplane image, a new method was given toovercome problems caused by sampling the whole Seriplane, Which are samplingand processing each pat of the whole Seriplane, synthesizing the result of eachpart by applying statistical method. Besides, the method and process of imageProcessing and feature extracteion were designed and optimized. This paper tried todevelop a system for identifying Seriplane image in a new environment──graphicos and visual programming environment, which is different from what was used byother domestic textile user. Meanwhile, a model for identification of yarn defectswas established and verified by the test, some new technical terms such asidentification threshold and neighboring yarn parts were pointed out.The digital image was captured by scanning standard photograph of cottonyarn. After the image was processed, two types of feature parameters wereextracted, they are statistical feature and direct visual feature.. The comparisonbetween computer method and Visual assessment show the following fact.Morphological method is suitable for processing Seriplane image. Imageprocessing method and algrithm designed in this paPer can realize the segmentbetween cotton yam and background. Statistical fethees can identify cotton yarnSeriplane of different grade, and can serve as feature parameters of grading cottonyarn Seriplane. The method for testing yam imperfection can recognize andclassify yam defects such as thick place, thin place and nep.The whole test resu1t is that: new method based on computer image processingcan objectively and effectively test Seriplane, which is reproducibility, eliminationof subjectivity and convenience When compared to traditional visual method. Theimage processing system, if further studied and imProved, promised to replacevisual assessment. Therefore this paPer have laid a useful theoretical and practicalfoundation for developmeni of an auomatic system for test yarn evenness. It alsomade a creative exploration for the aPplication of comPuter image processing intesting fiber material. In a word, the paPer is of importani practical value and greatsignificance.