高炉渣离心粒化图像颗粒分割研究

Study on Particle Segmentation of Blast Furnace Slag Centrifugal Granulation Image

高炉渣机械离心粒化过程中,基于图像识别的颗粒分割是实现粒径实时检测的关键因素。针对于高炉渣图像颗粒黏结而给测试带来困难的问题,在对图像做去噪处理后,结合改进的分水岭算法做颗粒分割,实现了黏结颗粒的准确分割。针对分水岭算法缺陷,结合积分图像的快速算法,对颗粒图像做局部阈值的二值化处理,并由形态学做细节弥补,采用了基于形态学重构距离图像算法以改进分水岭算法来分割颗粒图像,提高了处理效率,同时抑制了过分割现象,并准确地分割了黏结颗粒的图像,该算法为准确快速提取颗粒的特征参数奠定了基础。

In the process of centrifugal granulation of blast furnace slag,particle segmentation based on image recognition is a key factor for real-time particle size detection.Aiming at the difficulty of testing the granule slag image particle bonding problem,after denoising the image,the accurate segmentation of the bonded particles was realized by the improved watershed algorithm for particle segmentation.Aiming at the defects of watershed algorithm,combined with the fast algorithm of integral image,the particle image was binarized by local threshold,and the morphology was used to make up the details.The morphological reconstruction distance image algorithm was used to improve the watershed algorithm to segment the particle image.The processing efficiency is improved,the over-segmentation phenomenon is suppressed,and the bonded particle image is accurately segmented.The algorithm lays a foundation for accurately and quickly extracting the characteristic parameters of particle.