β Scribed by Giovanni Costantini; Daniele Casali; Renzo Perfetti
No coin nor oath required. For personal study only.
An analogic CNN algorithm is proposed for detection of multiple moving objects in high resolution, greyβscale images taken from a fixed camera. The algorithm, based on simple 3 Γ 3 templates, can be implemented using CNN hardware, providing the realβtime operation required in surveillance and traffic control applications. Efficient separation of moving objects from the background is obtained through automatic threshold selection. The performance of the proposed method is shown using realβlife indoor and outdoor video sequences. Copyright Β© 2004 John Wiley & Sons, Ltd.
π SIMILAR VOLUMES
Printed circuit board layout inspection methods are mostly based on local geometric information, therefore they are well suited to the CNN paradigm. Here the detection of two layout errors is considered namely, the breaks in the wires and some kind of short circuits. The designed analogic algorithms
This article presents a computational scheme for representing and interpreting the geometric tolerances (mainly size, orientation and position tolerances) assigned to polyhedral objects. The proposed representational scheme is based on a surface-based variational model. Variations are applied to a p