In single-view methods based on three-dimensional (3D) object geometry models in computer vision, a central problem is determining the correspondence of feature points in the model and the observed image. In this paper the author investigates the features of aspect identification methods used to solve this problem efficiently using matrices, then describes the causes of misidentifications. In addition, the author proposes a new identification matrix which improves this identification method statistically by using singular value decomposition, and general eigenvalues and eigenvectors, demonstrating the validity of the method using mathematical experiments. These improvements allow for a reduction of the computational burden for online identification of aspects while at the same time reducing the misidentification rate. This method can be an identification standard for norms of vectors and matrices. As a result, it is ideal for high-speed processing using hardware and for use in parallel systems.
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Information Systems
- Hardware and Architecture
- Computational Theory and Mathematics