|
Optimization Methods and Control Theory
Image processing toolkit inspired by mechanisms of human visual perception
A. P. Mashtakov, K. A. Putintseva Ailamazyan Program Systems Institute of Russian Academy of Sciences
Abstract:
We consider a problem of computer simulation of human vision. We develop software for image processing by methods based on the principles of visual perception by humans. The software Visual Processing performs the first stages of the visual signal processing by the human brain: blurring images (removing noise), edge detection (image contours) and orientation analysis (determining the angle of inclination of the contours). These steps have been modeled by Gaussian filter to blur and represent the image, a Gaussian derivative apparatus for edge detection, and a Gabor filter for determining the orientations. Visual Processing has been developed in C, tcl / tk using the libpng and libgsl libraries. It performs the following functions: Gaussian image blur; differentiation of the image using Gaussian derivatives; edge detection using the Laplacian of Gaussian (LoG filter); determining the orientation of the contours and lift of the image to the Lie group SE$_2$ using Gabor filters. Visual Processing is an open-source software that aims to build a platform for implementing and testing different image processing algorithms in the field of mathematical modeling of vision.
Key words and phrases:
image processing, vision model, human visual system, Gaussian derivatives, Gabor filter, roto-translation group.
Received: 26.11.2019 Accepted: 12.12.2019
Citation:
A. P. Mashtakov, K. A. Putintseva, “Image processing toolkit inspired by mechanisms of human visual perception”, Program Systems: Theory and Applications, 10:4 (2019), 111–139
Linking options:
https://www.mathnet.ru/eng/ps353 https://www.mathnet.ru/eng/ps/v10/i4/p111
|
Statistics & downloads: |
Abstract page: | 158 | Full-text PDF : | 42 | References: | 15 |
|