陳康++吳開華
摘要: 基于光截圖像的輪對(duì)在線檢測(cè)中,攝像機(jī)的標(biāo)定技術(shù)直接影響輪對(duì)參數(shù)的檢測(cè)精度。針對(duì)輪對(duì)在線檢測(cè)環(huán)境,設(shè)計(jì)了一套攝像機(jī)標(biāo)定系統(tǒng)。在該系統(tǒng)中,設(shè)計(jì)特制的平面標(biāo)定板,將其豎直固定于軌道面不同位置,通過異面特征點(diǎn)實(shí)現(xiàn)了攝像機(jī)的標(biāo)定。并利用攝像機(jī)標(biāo)定軟件實(shí)現(xiàn)了標(biāo)定板圖像采集,圖像處理,特征點(diǎn)計(jì)算機(jī)圖像坐標(biāo)提取和攝像機(jī)內(nèi)外部參數(shù)計(jì)算。實(shí)驗(yàn)結(jié)果表明,該標(biāo)定系統(tǒng)的攝像機(jī)標(biāo)定精度在0.1 mm以內(nèi),可應(yīng)用在輪對(duì)在線檢測(cè)中。
關(guān)鍵詞: 輪對(duì)在線檢測(cè); 攝像機(jī)標(biāo)定; 光截圖像; 異面特征點(diǎn)
中圖分類號(hào): TN 29文獻(xiàn)標(biāo)志碼: Adoi: 10.3969/j.issn.10055630.2016.01.001
Research on camera calibration technique of wheel set online
detection based on noncoplanar feature points
CHEN Kang, WU Kaihua
(College of Life Information Science and Instrument Engineering, Hangzhou Dianzi University,Hangzhou 310018,China)
Abstract: In the wheel set online detection based on structured light projection image technique, the camera calibration technique directly affects the precision of wheel set online detection. According to the environment of wheel set online detection, camera calibration system was designed. In the system, the special calibration plate was vertically installed on the rail surface. The noncoplanar feature points were used to realize the camera calibration. The camera calibration software realized the calibration plate image acquisition, the image processing and the extraction of computer image feature points. Finally, the software calculated internal and external parameters in the camera coordinate. The experimental results showed that the camera calibration accuracy was less than 0.1 mm. This system can be used on wheel set online detection.
Keywords: wheel set online detection; camera calibration; lightsection image; noncoplanar feature points
引言
輪對(duì)是車輛最重要的組成部件之一,其工作狀態(tài)對(duì)車輛的安全行駛有著重要影響[12]。隨著計(jì)算機(jī)技術(shù)、CCD傳感器技術(shù)等的發(fā)展,基于光截圖像測(cè)量的車輛輪對(duì)在線檢測(cè)技術(shù)成為目前輪對(duì)檢測(cè)領(lǐng)域研究的重點(diǎn)之一[35]。而在輪對(duì)在線檢測(cè)中,攝像機(jī)標(biāo)定技術(shù)又是一項(xiàng)關(guān)鍵技術(shù),標(biāo)定方法和精度對(duì)輪對(duì)在線檢測(cè)的最終結(jié)果有較大影響。
目前在輪對(duì)在線檢測(cè)中使用的攝像機(jī)標(biāo)定方法有如下幾種:文獻(xiàn)[6]將攝像機(jī)的內(nèi)部參數(shù)和外部參數(shù)分開求取,基于二維棋盤格標(biāo)定板,使攝像機(jī)在不同方位拍攝可移動(dòng)標(biāo)定板的多幅圖像完成攝像機(jī)內(nèi)部參數(shù)標(biāo)定,然后固定標(biāo)定板,采集一幅標(biāo)定板圖像,通過標(biāo)定板平面與對(duì)應(yīng)圖像的單應(yīng)性矩陣完成攝像機(jī)外部參數(shù)標(biāo)定,該檢測(cè)方法對(duì)于輪緣高度、寬度和垂直磨耗的測(cè)量重復(fù)性小于0.3 mm;文獻(xiàn)[7]提出將標(biāo)定板傾斜安裝在軌道面上,并使標(biāo)定板平面與激光面相重合,將三維空間標(biāo)定轉(zhuǎn)化為二維平面標(biāo)定,使用二元二次函數(shù)建立非線性畸變標(biāo)定模型,利用標(biāo)定板上特征點(diǎn)的像方坐標(biāo)和物方坐標(biāo)求出像方坐標(biāo)到物方坐標(biāo)的二元二次變換函數(shù),再利用該變換函數(shù)作用于拍攝的輪對(duì)圖像,實(shí)現(xiàn)系統(tǒng)的標(biāo)定,系統(tǒng)測(cè)量精度在0.5 mm以內(nèi);文獻(xiàn)[8]提出利用幾何光學(xué)成像原理,得到攝像機(jī)的理論位置和角度,設(shè)計(jì)出標(biāo)定板特征點(diǎn)理論成像坐標(biāo),調(diào)節(jié)攝像機(jī),使標(biāo)定板特征點(diǎn)實(shí)際成像坐標(biāo)與理論成像坐標(biāo)近似重合,再計(jì)算出攝像機(jī)實(shí)際位置與角度,此方法精度較高,但攝像機(jī)調(diào)節(jié)不便。
本文根據(jù)輪對(duì)在線檢測(cè)中實(shí)際環(huán)境要求,設(shè)計(jì)了一套攝像機(jī)標(biāo)定系統(tǒng)。并利用軟件集成很多算法,完成標(biāo)定板圖像采集,圖像處理,特征點(diǎn)計(jì)算機(jī)圖像坐標(biāo)提取和攝像機(jī)內(nèi)外參數(shù)計(jì)算。
光學(xué)儀器第38卷
第1期陳康,等:輪對(duì)在線檢測(cè)中基于異面特征點(diǎn)的攝像機(jī)標(biāo)定
1輪對(duì)在線檢測(cè)原理
圖1輪對(duì)在線檢測(cè)示意圖
Fig.1Wheel for online detection schemes輪對(duì)在線檢測(cè)原理如圖1所示。當(dāng)輪對(duì)進(jìn)入檢測(cè)區(qū)域時(shí),激光線光源以光平面的形式照射輪對(duì),在其表面形成一條光截曲線。該光截曲線包含了輪對(duì)的輪緣和踏面磨耗等輪對(duì)相關(guān)幾何尺寸參數(shù)信息。面陣CCD攝像機(jī)采集該輪對(duì)光截曲線圖像,通過圖像處理算法,將該光截曲線轉(zhuǎn)化為單像素光截曲線,根據(jù)攝像機(jī)標(biāo)定算法所求出的攝像機(jī)內(nèi)外部參數(shù)和光平面的角度,可計(jì)算出輪對(duì)光截曲線關(guān)鍵點(diǎn)的世界坐標(biāo),進(jìn)而求出輪對(duì)的輪緣厚度、高度和踏面磨耗等輪對(duì)相關(guān)參數(shù)。