Image Analysis Of Cotton Maturity In Longitudinal View

As a traditional natural cellulose fiber, cotton has an important position in textile fiber. Quality measurement of cotton fiber is significant to the production, supply and demanding of cotton textiles. Cotton maturity, the degree of development of fiber wall relative to its perimeter, is the first indicator in grade evaluation of cotton. Most of the physico-chemical properties of cotton fiber are affected by maturity, such as colour, strength, fineness, convolutions, moisture absorption and resilience and so on. Moreover, it is regarded as an important factor influencing textile performance and product quality. Hence, it is significant to measure cotton maturity rapidly, accurately and objectively.Existing measurement methods of cotton maturity are analyzed in this research, and are classified to direct methods and indirect methods. Usually, the accuracy of indirect methods is not very high due to the maturity is expressed based on quantity relations of middle variable. The direct methods such as lumen and wall thickness constract method, can extract parameters (wall thickness etc.) based on maturity definition. But production can not be guided well because their testing speed is slow and quantity of measured fibers is small. The studies in broad and at home show that image analysis is an effective way to measure maturity directly. Image analysis of cotton fiber can be divided into two broad categories:based on longitudinal images and based on cross-sectional images. Since the cross-sectional samples are difficult to obtained, the method based on cross-sectional images is not fit for rapid testing. Correspondingly, longitudinal samples are easy to get and information of fiber segments with a certain length can be supplied.A sample preparation scheme in dry air of cotton fiber is designed in this research. Fibers are cut to segments with the length of 0.5 mm by Y172 fiber microtome. The segments are dispersed in dry air. Images are acquired based on carpet automatic focus.Maturity measurement of cotton fiber based on longitudinal images is the recognition of cotton maturity actually. Firstly, boundary and stripe of fiber information is get used image processing techniques, then 3 feature parameters are extracted and pattern spaces of these parameters are built. The decision boundaries of different maturity classes are calculated used training samples. Image grayness and smoothness algorithms are used firstly to preprocess longitudinal images of cotton fiber. Fiber skeletons are exacted through image segmentation, thinning and intersected skeleton repairing algorithms. And then fiber stripes are obtained by edge detection and stripe quantization algorithms. After that, effective measured segments are selected.According to morphological feature of cotton fibers with different maturity, the parameters of average width Wf, convolution feature Sf and transparency feature Gf are selected and extracted to express maturity of cotton fiber.Based on fiber morphology of different cotton fibers in GB/T 6099-2008, an image library with 9 maturity classes is built as training samples. Pattern spaces are built used the feature parameters, and decision boundaries are get by spot probability density map of cotton fiber samples with different maturity classes. Then automatic classification of cotton maturity is realized.Average maturity, maturity distribution histogram and immature fiber content (IFC) can be obtained by the proposed method in this paper. Confirmatory experiments show that there is a linear correlation (R2 is greater than 0.85) between average maturity get from the proposed method and micronaire measured by High Volume Instrument (HVI).