DEVELOPMENT OF ALGORITHMS FOR INVARIANT THERMOVISION IMAGE RECOGNITION BY APPROXIMATED SINGULAR VALUE DECOMPOSITION

Igor Mokriš - Ľubomír Semančík

   In this paper a new method of application of Singular Value Decomposition (SVD) for invariant thermovision image recognition is presented based on the Approximated Singular Value Decomposition (ASVD). The analysis of singular values in dependence on the geometric transformations like translation, dilation and rotation of thermovision images is performed and algorithms for invariant image recognition are developed.

Keywords: approximated Singular Value Decomposition, invariant thermovision image recognition

[full-paper]