Object recognition with luminance, rotation and location invariance

Takami Satonaka; Takaaki Baba; Tatsuo Otsuki; Takao Chikamura; Teresa H. Meng

Object recognition with luminance, rotation and location invariance

Takami Satonaka^*, Takaaki Baba, Tatsuo Otsuki, Takao Chikamura, Teresa H. Meng

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

10 Citations (Scopus)

Abstract

In this paper we propose a neural network based on image synthesis, histogram adaptive quantization and the discrete cosine transformation (DCT) for object recognition with luminance, rotation and location invariance. An efficient representation of the invariant features is constructed using a three-dimensional memory structure. The performance of luminance and rotation invariance is illustrated by reduced error rates in face recognition. The error rate of using two-dimensional DCT is improved from 13.6% to 2.4% with the aid of the proposed image synthesis procedure. The 2.4% error rate is better than all previously reported results using Karhunen-Loeve transform convolution networks and eigenface models. In using DCT, our approach also enjoys the additional advantage of greatly reduced computational complexity.

Original language	English
Title of host publication	IEEE International Conference on Image Processing
Place of Publication	Los Alamitos, CA, United States
Publisher	IEEE Comp Soc
Pages	336-339
Number of pages	4
Volume	3
Publication status	Published - 1997
Externally published	Yes
Event	Proceedings of the 1997 International Conference on Image Processing. Part 2 (of 3) - Santa Barbara, CA, USA Duration: 1997 Oct 26 → 1997 Oct 29

Other

Other	Proceedings of the 1997 International Conference on Image Processing. Part 2 (of 3)
City	Santa Barbara, CA, USA
Period	97/10/26 → 97/10/29

ASJC Scopus subject areas

Computer Vision and Pattern Recognition
Hardware and Architecture
Electrical and Electronic Engineering

Cite this

Satonaka, T, Baba, T, Otsuki, T, Chikamura, T & Meng, TH 1997, Object recognition with luminance, rotation and location invariance. in IEEE International Conference on Image Processing. vol. 3, IEEE Comp Soc, Los Alamitos, CA, United States, pp. 336-339, Proceedings of the 1997 International Conference on Image Processing. Part 2 (of 3), Santa Barbara, CA, USA, 97/10/26.

@inproceedings{df59b353954f4ee48d3a621a8e035d55,

title = "Object recognition with luminance, rotation and location invariance",

abstract = "In this paper we propose a neural network based on image synthesis, histogram adaptive quantization and the discrete cosine transformation (DCT) for object recognition with luminance, rotation and location invariance. An efficient representation of the invariant features is constructed using a three-dimensional memory structure. The performance of luminance and rotation invariance is illustrated by reduced error rates in face recognition. The error rate of using two-dimensional DCT is improved from 13.6% to 2.4% with the aid of the proposed image synthesis procedure. The 2.4% error rate is better than all previously reported results using Karhunen-Loeve transform convolution networks and eigenface models. In using DCT, our approach also enjoys the additional advantage of greatly reduced computational complexity.",

author = "Takami Satonaka and Takaaki Baba and Tatsuo Otsuki and Takao Chikamura and Meng, {Teresa H.}",

year = "1997",

language = "English",

volume = "3",

pages = "336--339",

booktitle = "IEEE International Conference on Image Processing",

publisher = "IEEE Comp Soc",

note = "Proceedings of the 1997 International Conference on Image Processing. Part 2 (of 3) ; Conference date: 26-10-1997 Through 29-10-1997",

}

TY - GEN

T1 - Object recognition with luminance, rotation and location invariance

AU - Satonaka, Takami

AU - Baba, Takaaki

AU - Otsuki, Tatsuo

AU - Chikamura, Takao

AU - Meng, Teresa H.

PY - 1997

Y1 - 1997

N2 - In this paper we propose a neural network based on image synthesis, histogram adaptive quantization and the discrete cosine transformation (DCT) for object recognition with luminance, rotation and location invariance. An efficient representation of the invariant features is constructed using a three-dimensional memory structure. The performance of luminance and rotation invariance is illustrated by reduced error rates in face recognition. The error rate of using two-dimensional DCT is improved from 13.6% to 2.4% with the aid of the proposed image synthesis procedure. The 2.4% error rate is better than all previously reported results using Karhunen-Loeve transform convolution networks and eigenface models. In using DCT, our approach also enjoys the additional advantage of greatly reduced computational complexity.

AB - In this paper we propose a neural network based on image synthesis, histogram adaptive quantization and the discrete cosine transformation (DCT) for object recognition with luminance, rotation and location invariance. An efficient representation of the invariant features is constructed using a three-dimensional memory structure. The performance of luminance and rotation invariance is illustrated by reduced error rates in face recognition. The error rate of using two-dimensional DCT is improved from 13.6% to 2.4% with the aid of the proposed image synthesis procedure. The 2.4% error rate is better than all previously reported results using Karhunen-Loeve transform convolution networks and eigenface models. In using DCT, our approach also enjoys the additional advantage of greatly reduced computational complexity.

UR - http://www.scopus.com/inward/record.url?scp=0031338382&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031338382&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0031338382

VL - 3

SP - 336

EP - 339

BT - IEEE International Conference on Image Processing

PB - IEEE Comp Soc

CY - Los Alamitos, CA, United States

T2 - Proceedings of the 1997 International Conference on Image Processing. Part 2 (of 3)

Y2 - 26 October 1997 through 29 October 1997

ER -

Object recognition with luminance, rotation and location invariance

Abstract

Other

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this