东南大学学报(自然科学版)
東南大學學報(自然科學版)
동남대학학보(자연과학판)
JOURNAL OF SOUTHEAST UNIVERSITY
2010年
2期
326-330
,共5页
光学测量%图像处理%边界检测%机器视觉
光學測量%圖像處理%邊界檢測%機器視覺
광학측량%도상처리%변계검측%궤기시각
optical measurement%image processing%edge detection%machine vision
针对光学测量中圆形标志点面积小、背景不固定、投影畸变不一致等问题,提出一种基于区域检测和边界验证的圆点检测方法.在图像预处理后,使用边缘算子对图像进行区域检测,根据图像的连续特征,在边缘延展的基础上对所有可能的疑似圆点进行检测,在复杂背景中突出圆形标志点的特征,从而检测到变形的圆点;对得到的变形圆点集合,由图像和三维点之间的关系得到圆点的位置,利用圆点边界信息进行自适应验证,最终得到准确的圆形标志点集合.通过对各种复杂背景下的真实图像进行检测实验,证明基于区域检测和边界验证的检测算法简单、高效、可靠,可以在不同环境下快速准确地定位待检测点,有效地解决了标志点在各种情况下变形程度不同难以辨别的问题.
針對光學測量中圓形標誌點麵積小、揹景不固定、投影畸變不一緻等問題,提齣一種基于區域檢測和邊界驗證的圓點檢測方法.在圖像預處理後,使用邊緣算子對圖像進行區域檢測,根據圖像的連續特徵,在邊緣延展的基礎上對所有可能的疑似圓點進行檢測,在複雜揹景中突齣圓形標誌點的特徵,從而檢測到變形的圓點;對得到的變形圓點集閤,由圖像和三維點之間的關繫得到圓點的位置,利用圓點邊界信息進行自適應驗證,最終得到準確的圓形標誌點集閤.通過對各種複雜揹景下的真實圖像進行檢測實驗,證明基于區域檢測和邊界驗證的檢測算法簡單、高效、可靠,可以在不同環境下快速準確地定位待檢測點,有效地解決瞭標誌點在各種情況下變形程度不同難以辨彆的問題.
침대광학측량중원형표지점면적소、배경불고정、투영기변불일치등문제,제출일충기우구역검측화변계험증적원점검측방법.재도상예처리후,사용변연산자대도상진행구역검측,근거도상적련속특정,재변연연전적기출상대소유가능적의사원점진행검측,재복잡배경중돌출원형표지점적특정,종이검측도변형적원점;대득도적변형원점집합,유도상화삼유점지간적관계득도원점적위치,이용원점변계신식진행자괄응험증,최종득도준학적원형표지점집합.통과대각충복잡배경하적진실도상진행검측실험,증명기우구역검측화변계험증적검측산법간단、고효、가고,가이재불동배경하쾌속준학지정위대검측점,유효지해결료표지점재각충정황하변형정도불동난이변별적문제.
To solve the problems of circle detection in optical measurement,such as small size,unfixed background,and projection distortion inconsistencies,a general circle point detection method is presented,which is based on region detection and border verification.After an image is preprocessed,edge detector is applied to detect candidate circle points that have continuous edges,which can highlight the characteristics of circle points under complex background,and thus the deformed circle points can be detected.Then the position of every deformed circle point is calculated by the relationship between the image points and the corresponding three-dimensional points.Furthermore,using adaptive authentication,circle points can be testified according to the edge information.Finally,the true circle points can be obtained accurately.The method is carried out on various real images under complex background.The experimental results show that this new method is effective and credible,which can solve the identification problem caused by different degrees of deformation under various conditions.The points can be detected using the proposed method with high accuracy and speed in different environments.