【摘要】：機器視覺技術(shù)具有強大的穩(wěn)定性、廣泛適用的功能、超強的速度和非接觸特點。目前該技術(shù)在車輛尺寸測量、逆向工程、機械零部件的質(zhì)量檢測和特征形狀識別等方面得到了廣泛的應用。在這個科技迅猛發(fā)展和人工智能日漸進步的新時代,車輛的數(shù)量逐年成倍地增加,對交通造成了嚴重的影響。為了滿足當代交通智能化管理的需求、加大改裝車輛的法律監(jiān)管力度和保障社會的道路安全。本文利用機器視覺技術(shù),針對智能交通中車輛超載和超限問題,對車輛尺寸測量進行了針對性的研究,著重對測量系統(tǒng)的結(jié)構(gòu)設(shè)計與軟件算法進行研究,研究了尺寸參數(shù)進行非接觸性測量和提取的方法。論文主要內(nèi)容如下:首先對攝像機標定的方法進行了細致的和深入的研究,在針對一些常用的方法分析和比較后采用了一種基于立體環(huán)境視覺下的Open CV攝像機模型的標定方法。通過實驗,驗證了提出的Open CV攝像機模型標定方法的可行性,并對一些車輛的具體參數(shù)進行了標定。其次,采用了背景差分原理以實現(xiàn)對具有復雜背景的汽車圖像的準確定位。同時針對機器視覺圖像處理中的Canny圖像邊緣檢測算法出現(xiàn)邊緣模糊和漏檢的問題,所以提出圖像分塊理論和檢測邊緣算法,以修正原有的Canny邊緣檢測算法的不足。論文研究了在車輛的尺寸參數(shù)測量系統(tǒng)中使用的特征提取算法,對基于Hough變換的檢測直線、圓和Harris算子的角點理論模型進行了詳細研究。此外,將特征點匹配技術(shù)與區(qū)域圖像匹配技術(shù)結(jié)合起來,實現(xiàn)了立體匹配和自動檢測;研究了基于車輛測量系統(tǒng)的設(shè)計方法,通過分析系統(tǒng)功能開發(fā)了一種人機友好的上位機管理界面。
[Abstract]:Machine vision technology has strong stability, widely applicable functions, super speed and non-contact characteristics. At present, this technology has been widely used in vehicle dimension measurement, reverse engineering, quality detection and feature shape recognition of mechanical parts. In this new era with the rapid development of science and technology and the increasing progress of artificial intelligence, the number of vehicles has multiplied year by year, which has caused serious impact on traffic. In order to meet the demand of modern traffic intelligent management, the legal supervision of refitted vehicles should be strengthened and the road safety of the society should be guaranteed. In this paper, aiming at the problem of overloading and overloading of vehicles in intelligent transportation, this paper makes a targeted study on the measurement of vehicle size, focusing on the structure design and software algorithm of the measurement system. The method of non-contact measurement and extraction of dimension parameters was studied. The main contents of this paper are as follows: firstly, the method of camera calibration is studied in detail and deeply. After analyzing and comparing some commonly used methods, a calibration method of Open CV camera model based on stereo vision is proposed. The feasibility of the proposed Open CV camera model calibration method is verified by experiments, and some vehicle parameters are calibrated. Secondly, the background difference principle is adopted to accurately locate the automobile image with complex background. At the same time, aiming at the problem of edge blur and miss detection in Canny image edge detection algorithm in machine vision image processing, the theory of image segmentation and edge detection algorithm are proposed to correct the shortcomings of the original Canny edge detection algorithm. In this paper, the feature extraction algorithm used in vehicle dimension parameter measurement system is studied, and the corner theory model of detecting straight line, circle and Harris operator based on Hough transform is studied in detail. In addition, the feature point matching technology and the region image matching technology are combined to realize stereo matching and automatic detection, and the design method based on vehicle measurement system is studied. By analyzing the functions of the system, a man-machine friendly upper computer management interface is developed.
【學位授予單位】：哈爾濱理工大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：U495;TP391.41

【相似文獻】

相關(guān)期刊論文 前10條

1 ;線性尺寸測量工具[J];工具技術(shù);1976年02期

2 汪麟祥;戚元燧;;偏心影象剪切技術(shù)在微細尺寸測量中的應用[J];應用光學;1983年06期

3 В.Т.Макаренко;吳景文;;氣動式尺寸測量裝置[J];國外計量;1982年04期

4 荊旭東;;燕尾大端尺寸測量工具[J];機械工人.冷加工;1982年03期

5 李偉民,張巖,金如山;稠液霧對激光散射液滴尺寸測量影響的試驗研究[J];航空動力學報;1990年02期

6 橫X\R，

本文編號：2284223