Computer Vision Algorithms for Recognizing Basic Primitives of Objects Based on Digital Filters

The implementation of algorithms for primary image processing using one-dimensional digital filters adapted for DSP is considered. Methods of replacing a high bit operands with a lower one are used to increase performance, index arithmetic and methods using identifiers of graphic objects, replacing nonlinear operations with binary ones are presented. Algorithms for the detection of lines from the image, paths based on lines that form open or closed contours have been developed. The efficiency is shown when implemented on general-purpose processors.