Discrete-time signal processing
Discrete-time signal processing
Degraded Image Analysis: An Invariant Approach
IEEE Transactions on Pattern Analysis and Machine Intelligence
Invariants to Convolution in Arbitrary Dimensions
Journal of Mathematical Imaging and Vision
Digital Image Processing
Digital Image Restoration
Moment Forms Invariant to Rotation and Blur in Arbitrary Number of Dimensions
IEEE Transactions on Pattern Analysis and Machine Intelligence
Invariants for Recognition of Degraded 1-D Digital Signals
ICPR '96 Proceedings of the 13th International Conference on Pattern Recognition - Volume 2
Blurred image recognition by Legendre moment invariants
IEEE Transactions on Image Processing
Blur invariants: A novel representation in the wavelet domain
Pattern Recognition
Motion blur concealment of digital video using invariant features
ACIVS'06 Proceedings of the 8th international conference on Advanced Concepts For Intelligent Vision Systems
| Hi-index | 0.10 |
The blur invariant features developed by Flusser et al. [Proc. Internat. Conf. of Pattern Recognition, 1996, p. 389; IEEE Trans. Pattern Anal. Machine Intell. 20(6) (1998) 590; J. Math. Imaging Vision 13 (2000) 101] allow for exact pattern recognition between two continuous, infinite-extent signals related by a shift-invariant, centrosymmetric and energy-preserving filter. This paper addresses the case of discrete, finite-extent signals under these same filter constraints by establishing moment relations using a finite-extent convolution model. These relations demonstrate that the features of Flusser et al., while reporting good results, are not truly invariant to typical, discrete finite-extent signals. Using the established relations it is shown that, for Flusser et al.'s features to be truly invariant, the observed blurry signal must exactly result from the linear convolution of its associated reference signal and the shift-invariant filter--a condition that is likely not met in practice.