Higher-Order Neural Networks Applied to 2D and 3D Object Recognition

A higher-order neural network (HONN) can be designed to be invariant to geometric transformations such as scale, translation, and in-plane rotation. Invariances are built directly into the architecture of a HONN and do not need to be learned. Thus, for 2D object recognition, the network needs to be trained on just one view of each object class, not numerous scaled, translated, and rotated views. Because the 2D object recognition task is a component of the 3D object recognition task, built-in 2D invariance also decreases the size of the training set required for 3D object recognition. We present results for 2D object recognition both in simulation and within a robotic vision experiment and for 3D object recognition in simulation. We also compare our method to other approaches and show that HONNs have distinct advantages for position, scale, and rotation-invariant object recognition.The major drawback of HONNs is that the size of the input field is limited due to the memory required for the large number of interconnections in a fully connected network. We present partial connectivity strategies and a coarse-coding technique for overcoming this limitation and increasing the input field to that required by practical object recognition problems.

[1]  Geoffrey E. Hinton,et al.  Learning internal representations by error propagation , 1986 .

[2]  Wen-Thong Chang,et al.  Parameter Estimation And In-Plane Distortion Invariant Chord Processing , 1985, Other Conferences.

[3]  J D Downie,et al.  Experimental verification of modified synthetic discriminant function filters for rotation invariance. , 1990, Applied optics.

[4]  S. K. Rogers,et al.  The use of neural networks in PSRI target recognition , 1988, IEEE 1988 International Conference on Neural Networks.

[5]  Ming-Kuei Hu,et al.  Visual pattern recognition by moment invariants , 1962, IRE Trans. Inf. Theory.

[6]  L. Spirkovska,et al.  An empirical comparison of ID3 and HONNs for distortion invariant object recognition , 1990, [1990] Proceedings of the 2nd International IEEE Conference on Tools for Artificial Intelligence.

[7]  James L. McClelland,et al.  Parallel distributed processing: explorations in the microstructure of cognition, vol. 1: foundations , 1986 .

[8]  C. Lee Giles,et al.  Encoding Geometric Invariances in Higher-Order Neural Networks , 1987, NIPS.

[9]  H H Arsenault,et al.  Rotation-invariant digital pattern recognition using circular harmonic expansion. , 1982, Applied optics.

[10]  R. Haberman.  Elementary Applied Partial Differential Equations With Fourier Series and Boundary Value Problems , 1983 .

[11]  W. Pitts,et al.  How we know universals; the perception of auditory and visual forms. , 1947, The Bulletin of mathematical biophysics.

[12]  John Sullins.  Value Cell Encoding Strategies. , 1985 .

[13]  D A Jared,et al.  Inclusion of filter modulation in synthetic-discriminant-function construction. , 1989, Applied optics.

[14]  J. Ross Quinlan,et al.  Induction of Decision Trees , 1986, Machine Learning.

[15]  Colin Giles,et al.  Learning, invariance, and generalization in high-order neural networks. , 1987, Applied optics.

[16]  Charles R. Giardina,et al.  Elliptic Fourier features of a closed contour , 1982, Comput. Graph. Image Process..

[17]  John D. Downie,et al.  Determining Object Orientation with a Hierarchical Database of Binary Synthetic Discriminant Function Filters ( OPTICAL COMPUTING 1) , 1990 .

[18]  L. Spirkovska,et al.  Rapid training of higher-order neural networks for invariant pattern recognition , 1989, International 1989 Joint Conference on Neural Networks.

[19]  Zen Chen,et al.  Computer vision for robust 3D aircraft recognition with fast library search , 1991, Pattern Recognit..