Application of high-speed recursively-separable filters for processing noisy images

Authors: Tolmacheva A. S., Kamenskiy A. V., Tislenko A. A., Titov D. V.

Annotation: The paper provides a description of the algorithms developed for the implementation of high-speed filters, using the example of a moving average filter and a trapezoidal recursively separable filter. The speed evaluation results of the developed algorithms with an aperture dimension of 7x7 elements are presented in comparison with the classical twodimensional convolution. The influence of image processing formed by active-pulse television-measuring systems on the distance accuracy to objects of interest is estimated. As part of the study, the central element of the masks of the developed filters was changed in the range from -14 to 14, in parallel with this, the signal-to-noise ratio for the processed images was controlled.

Keywords: image processing, range measurement, trapezoidal recursive-separable filter, moving average filter, smoothing filters, signal-to-noise ratio, fast algorithms