基于FPGA的压力表盘机器视觉研究与实现

发布时间：2018-06-27 13:15

本文选题：球磨机 + 润滑油压　；参考：《昆明理工大学》2015年硕士论文

【摘要】：我国是一个矿产资源非常丰富的大国,储存的资源种类共有158种,而且分布非常广泛,有许多矿物质需要先进行破碎处理才能进行精选,大型球磨机为该生产工艺关键设备。选矿厂球磨机由于润滑油压力的不稳定所导致的烧瓦事故时有发生,因此球磨机润滑油压力是一个至关重要的实时监测参数。本文针对润滑油压力检测的可靠性问题,分析了球磨机静压润滑系统研究的现状和背景,给出了图像处理即基于FPGA的机器视觉表盘检测方法。该方法能够对压力表盘值进行机器识别,为压力传感器和控制系统提供冗余检测信号,可以降低油压过低而导致的烧瓦事故,提升油压检测系统的智能化程度。系统采用CCD就地摄像头获取压力表盘视频流,再由SAA7113视频采集芯片逐帧冻结图像传输到处理器FPGA中进行处理。针对摄像头采集表盘图像时出现畸变现象,采用插值法对图像进行校正,为后续的细节处理提供较好的前提条件。同时采用DoG滤波器、Marr-Hildreth边缘检测器、Hough变换和OPTA算法对表盘图像进行滤波、分割、提取以及数字细化。并给出一种改进的横竖端点寻点法(HVPP)和改进的模板匹配算法对提取的数字进行识别,能够实现压力表盘图像的智能读数。
[Abstract]:China is a large country with rich mineral resources. There are 158 kinds of resources stored and they are widely distributed. Many minerals need to be crushed before they can be selected. The large ball mill is the key equipment of the production process. The oil pressure of ball mill is a very important real time monitoring parameter because of the accident of burning tile caused by the instability of lubricating oil pressure in concentrator. Aiming at the reliability of lubricating oil pressure detection, this paper analyzes the present situation and background of the hydrostatic lubrication system of ball mill, and presents the image processing method based on FPGA for machine vision dial detection. The method can identify the pressure gauge disc value by machine, provide redundant detection signal for pressure sensor and control system, reduce the burning-off accident caused by too low oil pressure, and enhance the intelligent degree of oil pressure detection system. The system uses CCD camera to obtain the video stream of pressure gauge disk, and then transfers the frozen image frame by SAA7113 video acquisition chip to the processor FPGA for processing. Aiming at the distortion phenomenon when the camera collects the dial image, the interpolation method is used to correct the image, which provides a good precondition for the subsequent detail processing. At the same time, DoG filter, Marr-Hildreth edge detector, Hough transform and OPTA algorithm are used to filter, segment, extract and digital thinning the dial image. An improved Vertical Point finding method (HVPP) and an improved template matching algorithm are presented to recognize the extracted digits, which can realize the intelligent reading of the pressure gauge image.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TD453

【参考文献】