Theoretic design of a smart vision sensor

Abstract

The correlation method for image analysis is applied to the theoretic design of a smart vision sensor, which can be implemented to find the 2-dimensional position of static targets. In particular, the problem of finding the deviations along both X and Y directions is formulated as a matching process between the saved template of the desired position and the pictures of real static targets captured by a 'vision' element (for example, a CCD camera) through correlation analysis of the spatial shifts in 2-dimension. It is noted that the size of the captured images by the vision element can be reduced in our design in order to accommodate fast real-time application with reasonable accuracy and reliability. An example shows that our design idea exhibits a good performance in finding the deviations along both X and Y axes of a sheet of foam barriers and can be potentially used to design a real high performance and cost saving smart vision sensor.