棉花中白色异性纤维的激光成像快速检测方法

现有的棉花异性纤维分拣机采用可见光照明成像,难以识别棉花中混杂的白色异性纤维。为了有效地检测出与棉花颜色相同或相近的白色异性纤维,该文采用线激光照明成像的方法,在固定线激光功率、波长和相机光圈的条件下,利用不同的曝光时间,获取了12种典型的白色异性纤维与皮棉的图像,分析其成像曝光时间与图像对比度之间的关系,发现不同的关系曲线存在一段共有的最优曝光峰值时间。在此曝光时间内,同一图像中的白色异性纤维已经"过曝",而棉花还处于欠饱和状态。二者图像灰度值的明显差异可用于检测棉花中的白色异性纤维。该文在线激光功率为0.8 W,波长650 nm,相机光圈为8C,曝光时间为1.6 ms的条件下获取了1 500幅图像。试验表明,采用简单的固定阈值法,皮棉中12种典型的白色异性纤维的识别率达到了95.8%。研究结果为提高棉花中白色异性纤维的识别率和速度提供了一条新途径。

Laser imaging method for fast detecting white foreign fibers in cotton

Abstract: Foreign fibers, such as polypropylene baler twine, packaging materials, ropes, small pieces of cloths, nylon, and other different synthetic fibers, are almost invariably present in cotton during the cotton picking, storing, drying, and transporting. Their presence in textile processing may cause a disastrous quality problem in the textile industry because their physical and chemical properties are different from cotton. In recent years, machine vision systems have been used for sorting out the foreign fibers from cotton, however, all of the sorting machines cannot efficiently identify white foreign fibers in cotton, because their color is the same or very similar under the illumination of visible light. The laser imaging method previously suggested by our team can distinguish most of the white foreign fibers from cotton based on the difference of the microstructure of the sample surface, but the successful detection rate was not high and the speed of the detection was slow. Because the method was based on the calculation of the target size and distribution, or in the image acquisition, the laser light must project from some special angles of the sample surface for the camera can obtain the reflected light from the white foreign fibers. In this study, the image contrast of the white foreign fibers and the cotton background was first quantized by the calculation of the average pixel value in a frame of the laser image. Then, under the illumination of a line laser at a fixed power and wavelength, the images of 12 kinds of typical white foreign fibers on the cotton surface were obtained by a camera with a fixed aperture with different lengths of exposure time. Finally, the relationship between the lengths of different imaging exposure time and the image contrast was analyzed. It was found that there was a shared wave crest in all of the relationship curves and imaging in the exposure time of the wave crest in the same frame image. The gray level of the white foreign fibers had reached a saturated status, but the gray level of cotton was still unsaturated. Thus, using an optimized exposure time, the difference of the gray level of the laser images can be used for the detection of the white foreign fibers in cotton, including white paper, white plastic film, white semitransparent plastic mulch, white cloth, white density foam plastic, white nylon cord, white cotton string, white plastic cord, white plastic cardboard, white semitransparent polypropylene bags, white polyethylene foamed sheets, and white feathers. The experimental results indicated that, by a simple binary segmentation algorithm, using the images obtained under the illumination of the line laser at the wavelength of 650 nm, a power of 0.8 W, and by the camera aperture of 8C with an exposure time of 1.6 ms, the 12 kinds of white foreign fibers can be easily distinguished from cotton. In Matlab, the processing time of a frame image was less than 3 ms, and the successful detecting rate was up to 95.8 percent. In conclusion, a fast imaging method based on line laser and optimized exposing time was presented in the study, which can be used for on-line detection of white foreign fibers in cotton.