Akademik Veri Yönetim Sistemi
9th WSEAS International Conference on Signal Processing, Computational Geometry and Artificial Vision, Moscow, Rusya, 20 - 22 Ağustos 2009, ss.133-134
High speed machines and mechanisms are often studied from a sequence of images obtained from high speed videography. The use of markers printed or attached on moving parts can greatly assist in tracking the moving parts. We present the design of a marker suitable for planar motion analysis of mechanism. The marker is designed to make the task of image processing computationally less intensive so as to make real time motion analysis practical. Rosenfeld equivalence table algorithm is used to segment the input image. The new geometry of marker facilitates automatic tracking and provides both position and orientation information. Hu invariant moments are used for recognition of the marker shape in the segmented image. Markers are uniquely identified on the basis of a text code that is placed in a designated location with respect to the marker geometry. In this method, the bounding box for the text code is computed. Knowing the orientation of the marker and therefore the text orientation, it is possible to transform the sub-image, containing the text, so that the text is aligned horizontally. This will permit a standard OCR routine to read the text code. The motion of various moving parts in image sequence can be easily inferred once the position and orientation of each of the marker is known.