High-entropy signature using optimal coding
Jesse Sheng Jin · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1994
This paper presents a high entropy signature for matching sequence images. Many signatures have been developed for registration, such as Fourier descriptor, mathematic moments, generalized Hough transform, and parametric curves. However, some common problems exist, such as high computational cost, poor self-correlation, and low entropy. We propose a signature using the smallest enclosing circle of the pattern. Translation, rotation, and scaling invariants are obtained from the properties of the smallest enclosing circle. Optimal coding is applied to achieve a higher entropy than many other signatures. The signature is constructed into a hierarchical structure. Sequential techniques can be applied in registration.