S. Kroner, H. Schulz-Mirbach Fast adaptive calculation of invariant features. Tagungsband Mustererkennung  Home/Search Document Not in Database Summary Related Articles Check |
This paper is cited in the following contexts:
Invariant Features for Gray Scale Images - Schulz-Mirbach (1995) (13 citations) (Correct)
....that the features should be optimal with respect to a given (application specific) criterion. The question is how to choose the function f so that the group average A[f ] is optimal. Some ideas in this direction concerning the adaptive feature calculation are developed in the companion paper [3]. ....
S. Kroner, H. Schulz-Mirbach Fast adaptive calculation of invariant features. Tagungsband Mustererkennung
Index Editor's message - Parallel Image (Correct)
....it is possible to utilize the Volterra theory of nonlinear systems. We have shown that there are several links between these two approaches [7] An increase in the adaptivity of the features can be achieved by a parameterized calculation. Inspired by neural network techniques we have suggested in [17] a feedforward network which resembles with its binarytree like structure a subgraph of the signal flow graph of the class CT of cyclic shift invariant transforms. The network size is determined by the number of input elements. It consists of n = ld N layers which are sparsely connected with a ....
S. Kroner, H. Schulz-Mirbach Fast adaptive calculation of invariant features. In: G. Sagerer, S. Posch, F. Kummert (Hrsg.), Tagungsband Mustererkennung 1995 (17. DAGM Symposium), Reihe Informatik aktuell, S. 23-35, Springer 1995.
Fast Adaptive Calculation of Invariant Features - Kröner, Schulz-Mirbach Self-citation (Kroner) (Correct)
.... consider a simple example for the action of the group G T of cyclic translations on the four dimensional space V = C (cf. equation (1) with N = 4) For f(m) m[0] Delta m[2] we get the local function f l (m[i] m[i] Delta m[i 2] and the invariant feature A[f] m) m[0] Delta m[3] m[2] Delta m[1] The overall summation of the results of the local computations is most efficiently done by traversing a binary tree. This strategy for calculating invariant features is summarized in Figure 1. 3 How to incorporate adaptivity The invariant integral (4) permits the ....
....on each layer in the G T network whereas in the signal flow graph of the class CT only two different functions are used and the dimension of the input vector is maintained through all layers of the network. The invariance properties of G T networks are summarized in Theorem 1 which is proven in [3]. Theorem 1. Let n 2 IN and a G T network be given. We consider for N = 2 the pattern space V = C and the group GT of cyclic translations acting on V . Then the output of the G T network for a pattern m and every transformed pattern gm; g 2 G T , is identical. 4 How to construct ....
[Article contains additional citation context not shown here]
S. Kroner, H. Schulz-Mirbach Fast adaptive calculation of invariant features. Interner Bericht 1/95, Technische Informatik I, Technische Universitat HamburgHarburg, 1995.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC