Television images identification in the vision system basis on the mathematical apparatus of cubic normalized B-splines

Abstract

The solution the task of television image identification is used in industry when creating autonomous robots and systems of technical vision. A similar problem also arises in the development of image analysis systems to function under the influence of various interfering factors in complex observation conditions complicated the registration process and existing when priori information is absent, in background noise type. One of the most important operators is the contour selection operator. Methods and algorithms of processing information from image sensors must take into account the different character of noise associated with images and signals registration. The solution of the task of isolating contours, and in fact of digital differentiation of two-dimensional signals registered against a different character of background noise, is far from trivial. This is due to the fact that such task is incorrect. In modern information systems, methods of numerical differentiation or masks are usually used to solve the task of isolating contours. The paper considers a new method of differentiating measurement results against a noise background using the modern mathematical apparatus of cubic smoothing B-splines. The new high-precision method of digital differentiation of signals using splines is proposed for the first time, without using standard numerical differentiation procedures, to calculate the values of the derivatives with high accuracy. In fact, a method has been developed for calculating the image gradient module using spline differentiation. The method, as proved by experimental studies, and computational experiments has higher noise immunity than algorithms based on standard differentiation procedures using masks.

Keywords

• denoising
• spline
• approximation
• segmentation
• contouring