To determine a known object position and orientation within a scene is a task of fundamental importance. A scheme called reconstructive matching (RM), which is inherently invariant to shift, rotation, and optionally to scale changes is introduced. RM decomposes the matching within the source space into a set of mutually independent processes of dimensionality reduced by one. Applied to computerized tomography (CT), the approach allows CT object matching prior to their reconstruction.

Backprojection (BP) operation is of fundamental importance for image reconstruction from filtered projections in the framework of the computerized tomography (CT). The conventional BP algorithm, based upon the projection element spreading and cumulation along straight lines in the Cartesian coordinate image domain, provides a high quality reconstruction but suffers from being computationally expensive. Seeking a more effective BP algorithm, the authors have developed a new indirect scheme envisaging a two-stage process of reconstruction, in which the image polar coordinate domain is used as an intermediary between the projection and the image Cartesian coordinate domain. The BP scheme assumes a more complicated but more effective spreading of the projection elements. The algorithm developed provides the speedup of factor 3.3 with regard to the conventional BP algorithm. Experimental image reconstruction examples supplied by the indirect BP versus the conventional BP are given.>

The paper describes a new algorithm for computing the Hough/Radon transform (HT/RT) with reduced number of operations, and a dedicated single- or multiprocessor pipeline architecture, based on this algorithm. The latter envisages use of intermediate polar representation of image data. The implementation utilizes integer arithmetic only which makes the hardware realization compact and effective. The HT/RT architecture is equally suitable both for binary and gray-scale image transformation, which is achieved in constant time regardless of the scene contents. Since the transformation can be accomplished during a frame acquisition, the hardware device is able to act as a specialized HT/RT camera.>

Thresholding is a traditional tool for image regions classification based on their grey-level properties, especially when there are two kinds of homogeneous regions of intensity say, object and background. A new global thresholding scheme is described in this paper. Unlike the previously developed methods the approach proposed is independent of the object/background areas ratio, intensity transition slope and object/background shape; it is also resistant to noise.>