提高锡、钨、金等金属回收率的摇床分带图像分割法研究

摇床广泛应用于锡、钨、金等微细粒矿物的回收。摇床分带技术是实现摇床优化控制和提高选别指标的前提。目前，国内摇床分带主要由人工操作，分带精确度不高、实时性滞后、选别指标较低，故急需研究新的摇床分带技术。基于数字图像处理的摇床分带技术，通过图像分割可以精确自动地实现摇床分带，目前国内外针对摇床分带图像分割还没有提出相应的算法，因此本项目对基于萤火虫算法的摇床分带图像分割法进行研究。首先通过获取大量的现场摇床分带图像以及进行图像分析，掌握摇床分带的特征及规律；研究萤火虫算法的基础理论；研究步长、决策半径、萤火虫分布、萤火虫位置更新方式对萤火虫算法收敛性和实时性的影响；从摇床分带图像的单阈值和多阈值分割两方面出发，结合摇床分带的特征，提出分割精度和实时性较高的萤火虫算法；对分割算法进行编程、现场试验、分析比较和效果评价，获得摇床分带图像分割的新理论和新技术，以提高稀贵金属锡、钨、金等资源的回收率。

Shaking table is widely used in the recovery of tin、tungsten、gold and other micro-fine metal. The technology of shaking table zone is the premise to achieve the optimum control of table and improve separation indexes. At present, shaking table zone technology is by artificial at home, which has the lower accuracy, real-time and separation indexes, therefore the new technology is need to research urgently. The method based on digital image processing can achieve precise and automatic zoning by way of image segmentation, now the corresponding algorithm has not been presented at home and abroad for segmentation of shaking table zone image, in the project, so the segmentation method of shaking table zone image that based on glowworm swarm algorithm is studied. Firstly, master the characteristics and laws of zoning by obtaining a large of number of shaking table zone images on the scene and analyzing images; research the fundamental theory of glowworm swarm algorithm; study the influence on the convergence and real-time of glowworm swarm algorithm that caused by step-size, decision range, distribution of glowworms, and movement-phase; combined with the characteristics of zoning, put forward new algorithms that have higher segmentation accuracy and real-time, in the light of bi-level threshold and multilevel threshold of shaking table zone image; program on segmentation algorithm ,field experiments, analysis and evaluation of effects, then obtain new theories and new techniques for segmentation of shaking table zone image, to improve the resource recovery of tin、tungsten , gold and other rare metal.