The Automatic Detection In The Cross-section Of Fibers

With the development of living standard, the demand of chasing comfort and comeliness nowadays is increasing today; and many types of differential fibers are emerging continuously. Because there is a close relationship between the performance and the shape of the fiber cross-section, all kinds of profiled fibers should be detected automatically and in mass in the production of chemical fiber and textile. Among the means of detecting fibers, it is convenient to get the characteristic parameters by detecting the fiber cross-section. However, at present they are mostly detected by manual method in our country, then not only it is heavy in the work load and inefficient, but also the data is not accurate and lack of good instability. So it is quite necessary to study the automatic detection in cross-section of profiled fibers.In order to reduce the work load and improve the preciseness, the automatic detection system in cross-section of profiled fibers is firstly put forward in this thesis, whose 5 parts should be realized automatically, including the fiber sampling, fiber slicing, slice taking, automatic focusing and image processing. And the key technology of fiber slicing, automatic focusing and image processing parts are mainly studied in this thesis. And the non-hollow profiled fibers as the studied objects of this thesis were obtained from the former spinning experiment.In the part of fiber slicing, it is the first time to propose that the biological tissue slicer should be combined with the freezer which is a way that applied into the fiber slicing. And then three slice methods were used to do the slice experiment. After analyzing and the comparing, it showed that the new way had made great progress, for example the speed of fiber slicing was quite fast and the rate of success was high, which is propitious to the realization of the automatization for slice taking in the next step. Then main parameters were confirmed optimumly by numerous experiments, and they would be usefull to guide the production of fiber.In the part of auto-focusing, several sorts of common auto-focusing evaluation functions were discussed. And they were utilized into the auto-focusing simulation experiment respectively based on the maximum searching algorithm, and the result evidenced that the speed of auto-focusing was faster by the square gradient function (SGF), At the same time, the reason why multi-focal plane question appears, various ways was used to avoid the multi-focus and obtain satisfying results.In the part of image processing of this thesis, the various parameters in Canny operator were studied, which have an effect upon the edge detection of fiber cross-section, and an new assembled algorithm was brought forward which based upon the combination of the mathematical morphology method and the Canny operator. The experiment indicated that the new algorithm could detect the edge clearly and completely, and restrain noises perfectly. Moreover the characteristic parameters of these fibers were measured by using the new algorithm and programmes designed for the images of fiber cross-section in the thesis.The key technology of the automatic detection system has been resolved in this thesis, which achieved the purpose expected. The automatic detection system in cross-section of profiled fibers is firstly put forward in our country. It will provide a practical foundation to further study the inner relationship between the shape parameters of fiber and its performance. It can also offer the essential fact basis for improving the quality of the textile fibers and promoting the production of the chemical fiber, and the development of textile industry and clothing industry. Furthermore, the system can be applied into the detection of other fields.