Research On Evaluation Of Knitted Fabric Pilling Grade Based On Three-Dimensional Scanning Technoligy

In the knit textile industry, the pilling performance of fabric is one of the most importent indicators to reflect the quality of knitted fabric. Not only in fabric design but also in textile trade, measurement of knitted fabric pilling is essential. The traditional methods are directly observated by the tester, but when large quantities of test, it will demonstrate its inherent limitations, like consuming a lot of time and measurement results are largely influenced by the competent. In recent years to improve the traditional measure method, a lot of new methods have emerged to improve efficiency. There are two common, one is two-dimensional image processing method. It take photos of the test fabrics, then use certain methods to get the boundaries of pilings, and use these boundaries to obtain the various features of pilling, then evaluate fabric pilling grade; the other is using indirect method to obtain three-dimensional coordinates and other surface information of the test fabrics, which reflected the characteristics of fabric pilling values, according to these characteristics the grade of fabric pilling can be determined. Both methods are greatly extent changed the shortcoming of traditional methods, and make intelligence methods become popular. But they are not the best, the first method is limited by fabric color and texture when calculate the threshold; the second method also easy generate bad values because the front point information have great impact on the next one. In this paper, Xystum three-dimensional scanner can accurately obtain the information of the knit fabric surface, then use OpenGL simulation technology can reconstruction the fabric surface, make the detection process intelligent and visualization.In this paper, using Xysum three-dimensional scanner to get knit fabric surface coordinate information for intelligent evaluation of fabric pilling grade is being studied, also introduced how to use Xystum three-dimensional scanner to obtain the surface information of software material and discuss how to regularization scan data to achieve better result of modeling. Then achieve a whole visualization process of OpenGL-based simulation and information separation process. During the intelligent tradition process to get pilling values a new method named spaced net structured method is used, it can correct the fluctuation in the scanning position while made by mechanical vibration and laser scanning reflection angle offset. Then use the two-dimensional wavelet denosing algorithm to deal with the scanning height information to eliminate the effect of date noise caused by color, gloss change and long fiber. And then use the height average and uniform threshold method to obtain the threshold information. In the VC emulators using different colors can see the separate result very clear. After separate by threshold, Four Unicom Traversal method is used to separate every pill, Then the pilling values like average height, maximum height, projected area, volume, also average projected area, maximum projected area, total volume, average volume, maximum volume. After a number of test and comparison relationship between pill values and pilling grades is being built, which can be used to intelligent evaluat knit fabric pilling grade.In the paper data collection is using Xystum three-dimensional scanner to get, it has very high precision, and the minimum scan interval is about one micron. The data processing and analog display are accomplished easy, A modeling software named Playwork is been used for comparison, which will make the VC modeling process more accurate.In data optimization process, First use the Convert software to convert the scan data information to ASCII format for the computer can easy read, Then use VC program language to read the converted data into model, and then use spaced net structured method to correct the scan point offset in the XY plane, use median filter and discrete wavelet transform to deal with the height coordinate, which would effectively reduce the deviation of optical data caused by the impurities, filament fiber and machine vibrations.In the process of pilling separate, height average, uniform threshold, single back reduce automatic threshold and total back reduce automatic threshold methods were used, the reference are good, different fiber use different method the results were different, VC simulation technology can make operator see the result of separate clearly because of it's well visibility, So suitable threshold method can be chose. Also operator can set in his determine, which make subjective and objective effective combinated, effective enhance the detection of detection.In the paper threshold is used to separate pilling information and get pilling eigenvalues in intelligent evaluation of pilling grade process, feature vector can be built by these eigenvalues, then calculate the vector to the standard class feature vector's distance, the test fiber's pilling grade is the same as the standard pilling fiber with the lowest distance between them.