چكيده به لاتين
Clifford or Geometric Algebra is a powerful mathematical tool that has many applications in research fields like electromagnetic signal processing, image processing, machine vision, neural networks, and kinematics. Algorithms based on this geometric framework have a significant computational load that general-purpose processors are unable to handle them properly. Therefore efficient implementation of geometric algebra algorithms needs special purpose hardware architecture. In this project, an efficient hardware architecture designed for RGB color image edge detection in 3D Clifford algebra space. The color edge detection algorithm, called saturation gradient and geometric product, is generalized for operating on CMY color images. In addition to edge detection, this approach can be used for image processing on a specific color region.
The results from applying this algorithm compared to other approaches show its ability to find edges as well as best of them. With respect to processing goals, the algorithm parameters can be tuned by the user to achieve best results. The designed hardware is compared with two hardware implementations for Clifford color edge detection algorithms. In comparison to Gaalop pre-complier hardware, at the same frequency and same speed up, the proposed design used fewer logic resources. The edge detection algorithm execution on proposed hardware design shows x351 speed up compared to its execution on a general geometric algebra co-processor.
