Filaments Stratified Imaging Process And Its Application Study

The viscose filament industry of our country started at 1950s. But now we have won the first place in the output, and have made the huge progress. However, we still have a long way to go to catch up the overseas companies in the quality of viscose filaments. Because the technique of production of viscose filaments is not good enough. Our country's viscose filaments mostly cannot achieve the international standard, So we cannot get a good score in the foreign trade. The main reason why the quality of our filaments cann't get up to the international standard is the short of the number of the viscose filaments yarn which leading to the asymmetry of the cloth coloration. This is also the main reason why ours filaments are not good as the filaments produced in Japan and Europe. We can get the idea that measuring the number of the filaments yarn quickly and finding out the condition of the bushing tip can provide for us the important reference. This paper represents a image fussing method which based on the Optical Sectioning Tomography. This method can replace the man-counting method which is inefficient and low at the rate of accuracy. Tested by experiments, this method has the strong points of quick measuring, accuracy, simple structure and low cost. This method can do something in promoting the quality of viscose filaments.Machine vision and image processing have developed into consummate technology. And the two technology have been applied to every walk of life. This system combine machine vision and image processing skillfully. It is applied to viscose filaments measuring and counting, and it gets a good result. In this system, getting the images at different distance in focus is the premise. This system resolves the imaging problem by using the microscope's advantages in getting images of fine objects and choosing favourable source after theory computing and practice testing. The key of the system is fussing accurately of a sequence of viscose filaments images. Image Fusion Method Based on Point Spread Functions, image fusion methods based on mean gradient and image fusion algorithm based on wavelet transform have been applied in the fussing of filaments images. The results of fussing is judged by maximum entropy and RMSE. Comparing the outcomes of these three methods, we find that the fusion method based on wavelet transform is more suitable to the system.Image enhancement and image segmentation are needed to filter the noise, enhance the contrast by suppressing unconcerned information before and after image fussing of viscose filaments. This system effectively suppresses parts of noises by synthetically using image enhancement methods of spatial domain and frequency field. A new method is proposed on the basis of Otsu algorithm. in this method we segment image with local threshold and then joint the different parts to get one complete binary image.After fussing the viscose filament images occurs disconnection problems and blocking problems. In this paper, some classical morphological algorithms are used to solve these problems. These algorithms are erosion, dilation, opening, closing and hit-or-miss transform. Viscose filaments disconnection problems and blocking problems are effectively solved In this paper by using filament-grow method and thinning the filaments. The preliminary works are done for counting the filaments.Two method which are called connected domain tracking algorithms and connected domain departing algorithms are used to counting the filaments after a series processing. Comparing the counting result and actual filaments, we find that Measuring precision reach 100% and 96.8% which proved that the system is feasible and accurate.