Real-time Sub-pixel Cross Bar Position Metrology M any measurement application fields need to calculate cross bar intersection locations of horizontal and vertical bars. The system we developed and that we present in this paper is an embedded system that measures cross bar positions with sub-pixel accuracy on 1024  1024 pixel images delivered by a camera at a 50 MHz data rate in real time. This is done using an algorithm that looks for intersection areas and then locally calculates two lines representing horizontal and vertical bars. The two line intersection is considered to be the bar intersection. To achieve real time, we developed a hybrid architecture in which low level processes are implemented into FPGAs and others into DSPs. As a result, at the end of the camera scanning (20 ms), all calculations are completed and the results are available. # 2002 Published by Elsevier Science Ltd. D. Rivero 1 , M. Paindavoine 2, * and S. Petit 1 1 Thomson Tubes and Displays, Avenue du Ge´ne ´ral de Gaulle, 21110 Genlis, France. 2 Universite´ de Bourgogne, LE2I, Aile des Sciences de l’Inge ´nieur, 21011 Dijon, France. Introduction In several areas, like human movement analysis, or in our case TV metrology, a cross bar target is displayed and intersection positions are calculated to obtain information about human movements or about TV screen adjustments. This requires highly accurate measurements and most of the time, high-speed proces- sing for feedback effect. In our case, this means using a 1024  1024 pixel CCD camera with a 50MHz data rate. Using our system we then reach an accuracy of 50 mm in a 1m wide screen using a sub-pixel algorithm. Research has been carried out on real-time measure- ment systems as in astronomic applications [1], in object corner positioning [2] and in human movement analysis [3]. These video systems allow validation of the position measurement algorithm implementation at a maximum of 10 MHz pixel data rate. In order to process the 1024  1024 images in real time, we selected an algorithm which involves as a first step the object location in the image and as a second step a two dimensional object position measurement in this located area. The location step processes the whole image and is made using low-level image processing while the second step is achieved on all the pixels in the located areas. Thus step comprises centre of mass calculations and using these results a least mean square line regression is carried out to represent horizontal and vertical bars. The intersection of these two lines is considered as being the bar intersection. Thus, this algorithm requires both low level and high level processing. *E-mail: paindav@u-bourgogne.fr 1077-2014/02/$35.00 r 2002 Published by Elsevier Science Ltd. Real-Time Imaging 8, 105–113 (2002) doi:10.1006/rtim.2001.0259, available online at http://www.idealibrary.com on