基于图像处理的矿石粒度检测方法研究

发布时间：2018-09-13 09:15

【摘要】：矿石粒度是矿物加工工艺的一项重要指标,磨矿作业需要对矿石粒度进行检测,根据检测结果调整相应工艺参数。目前常采用筛分、沉降等常规方法进行粒度检测,这些检测方法耗时长、效率低,而且受检测人员的主观影响较大。针对以往检测方法的不足,本论文采用基于数字图像处理的检测方法,经过实践证明,这种检测方法能够快速、准确的对矿石粒度进行分析测量。本论文研究的主要内容有:1.对矿样原始图像进行预处理,首先对原始图像进行灰度化,接着对于目标矿粒与背景不易区分的现象进行了对比度调节,之后分析对比了两种典型的滤波算法,选取中值滤波算法滤除了图像噪声。2.对矿样图像进行分割。对不同的分割算法进行了分析与对比,最后选用了基于阈值的图像分割方法,成功的将目标矿粒与背景分离。3.矿样图像的形态学处理。对分割后的图像进行形态学处理,平滑图像噪声,填充由于矿粒反光而形成的孔洞。4.粘连矿粒的分割。在传统分水岭算法的基础上,分析对比了几种改进的分水岭算法分割效果,最后采用了基于标记符控制的分水岭分割算法,成功将粘连矿粒分离开。5.对处理后图像中的连通区域进行标记,计算每个连通域中的像素个数,通过比例换算,得到矿粒的实际粒度。6.完成软件编译,输出粒度分布曲线,将软件分析结果与筛分结果进行对比。通过实验证明,本论文的检测方法成功、有效的统计出了矿石的粒度分布,与传统方法相比,基于图像处理的检测方法具有操作简便、准确高效的优点,同时,对矿石粒度的在线检测也取得了很好的效果,该检测方法对提高磨矿作业效率以及推动选矿自动化的发展都具有重要意义。
[Abstract]:Ore particle size is an important index of mineral processing technology. Grinding operation needs to detect ore particle size and adjust the corresponding process parameters according to the test results. At present, conventional methods such as sieving and settling are often used to detect particle size. These methods are time-consuming, inefficient and subject to the subjective influence of the examiners. In view of the shortcomings of the previous detection methods, this paper adopts the detection method based on digital image processing. It has been proved by practice that this detection method can analyze and measure the ore particle size quickly and accurately. The main content of this thesis is 1: 1. Preprocessing the original image, first graying the original image, then adjusting the contrast between the target ore particles and the background, then analyzing and comparing two typical filtering algorithms. Select median filter algorithm to filter image noise. 2. The mineral image is segmented. The different segmentation algorithms are analyzed and compared. Finally, the threshold-based image segmentation method is used to separate the target ore particles from the background successfully. Morphological processing of mineral image. The segmented image is processed by morphology to smooth the noise of the image and fill the hole. 4. The division of mineral particles. Based on the traditional watershed algorithm, this paper analyzes and compares the segmentation effects of several improved watershed algorithms. Finally, the watershed segmentation algorithm based on marker control is used to separate the adhesion particles successfully. The connected region of the processed image is marked, the number of pixels in each connected domain is calculated, and the actual particle size of the ore is obtained by the scale conversion. The software is compiled, the granularity distribution curve is outputted, and the results of software analysis and screening are compared. It is proved by experiments that the detection method in this paper is successful, and the particle size distribution of ore is calculated effectively. Compared with the traditional method, the method based on image processing has the advantages of simple operation, accuracy and high efficiency, at the same time, The on-line detection of ore size has also achieved good results. This method is of great significance to improve the grinding efficiency and promote the development of mineral processing automation.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD91;TP391.41

【参考文献】