Multiaxial Haar-like feature and compact cascaded classifier for versatile recognition

Jun Nishimura, Tadahiro Kuroda

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

A versatile recognition algorithm has been proposed to process image, sound, and 3-D acceleration signals with a common framework at low calculation cost. Firstly, a novel 1-D Haar-like feature is used to roughly extract frequency information from temporal signals. Biaxial and mean-embedded Haar-like features are proposed to extract the standard deviation and the interaxial correlation from 3-D acceleration signals. Secondly, two techniques are proposed to build a compact cascaded classifier. Redundant feature selection (RFS) incorporates the features which are already selected in previous stage classifiers to reduce the calculation cost. A dynamic look-up table (DLUT) is proposed to construct a look-up table-based weak classifier with the smallest possible number of bins. A train loss function is by globally optimized using dynamic programming. The proposed algorithm is tested experimentally on speech/nonspeech classification and human activity recognition. The proposed algorithm yields a speech/nonspeech classification performance comparable to the state-of-art method called MFCC while reducing the calculation cost by 100 times. The algorithm also achieves human activity recognition accuracy of 96.1% with calculation cost reduction of 84% compared with the state-of-art method based on C4.5 decision-tree classifier using the basic statistical features. The proposed algorithm has been employed to build the versatile recognition processor.

Original languageEnglish
Article number5482111
Pages (from-to)1786-1795
Number of pages10
JournalIEEE Sensors Journal
Volume10
Issue number11
DOIs
Publication statusPublished - 2010

Fingerprint

classifiers
Classifiers
costs
Costs
dynamic programming
cost reduction
Bins
Decision trees
Cost reduction
Dynamic programming
central processing units
Feature extraction
standard deviation
Acoustic waves
acoustics

Keywords

  • Cascaded classifier
  • Haar-like feature
  • versatile recognition

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Instrumentation

Cite this

Multiaxial Haar-like feature and compact cascaded classifier for versatile recognition. / Nishimura, Jun; Kuroda, Tadahiro.

In: IEEE Sensors Journal, Vol. 10, No. 11, 5482111, 2010, p. 1786-1795.

Research output: Contribution to journalArticle

@article{f47515a897a743b8a8eb634b61b88972,
title = "Multiaxial Haar-like feature and compact cascaded classifier for versatile recognition",
abstract = "A versatile recognition algorithm has been proposed to process image, sound, and 3-D acceleration signals with a common framework at low calculation cost. Firstly, a novel 1-D Haar-like feature is used to roughly extract frequency information from temporal signals. Biaxial and mean-embedded Haar-like features are proposed to extract the standard deviation and the interaxial correlation from 3-D acceleration signals. Secondly, two techniques are proposed to build a compact cascaded classifier. Redundant feature selection (RFS) incorporates the features which are already selected in previous stage classifiers to reduce the calculation cost. A dynamic look-up table (DLUT) is proposed to construct a look-up table-based weak classifier with the smallest possible number of bins. A train loss function is by globally optimized using dynamic programming. The proposed algorithm is tested experimentally on speech/nonspeech classification and human activity recognition. The proposed algorithm yields a speech/nonspeech classification performance comparable to the state-of-art method called MFCC while reducing the calculation cost by 100 times. The algorithm also achieves human activity recognition accuracy of 96.1{\%} with calculation cost reduction of 84{\%} compared with the state-of-art method based on C4.5 decision-tree classifier using the basic statistical features. The proposed algorithm has been employed to build the versatile recognition processor.",
keywords = "Cascaded classifier, Haar-like feature, versatile recognition",
author = "Jun Nishimura and Tadahiro Kuroda",
year = "2010",
doi = "10.1109/JSEN.2010.2049740",
language = "English",
volume = "10",
pages = "1786--1795",
journal = "IEEE Sensors Journal",
issn = "1530-437X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "11",

}

TY - JOUR

T1 - Multiaxial Haar-like feature and compact cascaded classifier for versatile recognition

AU - Nishimura, Jun

AU - Kuroda, Tadahiro

PY - 2010

Y1 - 2010

N2 - A versatile recognition algorithm has been proposed to process image, sound, and 3-D acceleration signals with a common framework at low calculation cost. Firstly, a novel 1-D Haar-like feature is used to roughly extract frequency information from temporal signals. Biaxial and mean-embedded Haar-like features are proposed to extract the standard deviation and the interaxial correlation from 3-D acceleration signals. Secondly, two techniques are proposed to build a compact cascaded classifier. Redundant feature selection (RFS) incorporates the features which are already selected in previous stage classifiers to reduce the calculation cost. A dynamic look-up table (DLUT) is proposed to construct a look-up table-based weak classifier with the smallest possible number of bins. A train loss function is by globally optimized using dynamic programming. The proposed algorithm is tested experimentally on speech/nonspeech classification and human activity recognition. The proposed algorithm yields a speech/nonspeech classification performance comparable to the state-of-art method called MFCC while reducing the calculation cost by 100 times. The algorithm also achieves human activity recognition accuracy of 96.1% with calculation cost reduction of 84% compared with the state-of-art method based on C4.5 decision-tree classifier using the basic statistical features. The proposed algorithm has been employed to build the versatile recognition processor.

AB - A versatile recognition algorithm has been proposed to process image, sound, and 3-D acceleration signals with a common framework at low calculation cost. Firstly, a novel 1-D Haar-like feature is used to roughly extract frequency information from temporal signals. Biaxial and mean-embedded Haar-like features are proposed to extract the standard deviation and the interaxial correlation from 3-D acceleration signals. Secondly, two techniques are proposed to build a compact cascaded classifier. Redundant feature selection (RFS) incorporates the features which are already selected in previous stage classifiers to reduce the calculation cost. A dynamic look-up table (DLUT) is proposed to construct a look-up table-based weak classifier with the smallest possible number of bins. A train loss function is by globally optimized using dynamic programming. The proposed algorithm is tested experimentally on speech/nonspeech classification and human activity recognition. The proposed algorithm yields a speech/nonspeech classification performance comparable to the state-of-art method called MFCC while reducing the calculation cost by 100 times. The algorithm also achieves human activity recognition accuracy of 96.1% with calculation cost reduction of 84% compared with the state-of-art method based on C4.5 decision-tree classifier using the basic statistical features. The proposed algorithm has been employed to build the versatile recognition processor.

KW - Cascaded classifier

KW - Haar-like feature

KW - versatile recognition

UR - http://www.scopus.com/inward/record.url?scp=77956697123&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77956697123&partnerID=8YFLogxK

U2 - 10.1109/JSEN.2010.2049740

DO - 10.1109/JSEN.2010.2049740

M3 - Article

VL - 10

SP - 1786

EP - 1795

JO - IEEE Sensors Journal

JF - IEEE Sensors Journal

SN - 1530-437X

IS - 11

M1 - 5482111

ER -