Search Results

Now showing 1 - 3 of 3

Some of the metrics are blocked by your
consent settings
Invariant range image multi-pose face recognition using gradient face, membership matching score and 3-layer matching search
(2005-01-01)
Pansang, Seri
;
Attachoo, Boonwat
;
Kimpan, Chom
;
Sato, Makoto
The purpose of this paper is to present the novel technique to solve the recognition errors in invariant range image multi-pose face recognition. The scale, center and pose error problems were solved by using the geometric transform [13]. Range image face data (RIFD) was obtained from a laser range finder and was used in the model to generate multi-poses. Each pose data size was reduced by linear reduction. The reduced RIFD was transformed to the gradient face model for facial feature image extraction and also for matching using the Membership Matching Score model. Using this method, the results from the experiment are acceptable although the size of gradient face image data is quite small (659 elements). Three-Layer Matching Search was the algorithm designed to reduce the access timing to the most accurate and similar pose position. The proposed algorithm was tested using facial range images from 130 people with normal facial expressions and without eyeglasses. The results achieved the mean success rate of 95.67 percent of ±12 degrees up/down and left/right (UDLR) and 88.35 percent of ±24 degrees UDLR. Copyright © 2005 The Institute of Electronics, Information and Communication Engineers.
Some of the metrics are blocked by your
consent settings
3-D object recognition from shadow shape curves
(2000-12-01)
Pansang, Seri
;
Kimpan, Chom
The 3-D object recognition can be normally done by collecting 2-D projections taken at various angles viewing, which requires a large size of the database. In this paper we present the 3-D object-recognizing approach by analyzing the shadow shape curve of 2-D from horizontally 360° viewing angles, projections which depends on typical characters of objects. The process starts from shooting the subject with the light source behind the subject or a subject with a light background while the subject is being turned around in various viewing angles. The process is independent on the size of the shadow shape curves or position of the subject taken. This process shows the shadow shape curves which are then converted to patterns of slope scores and later matched with the standard slope score patterns in database. By comparing the sequent aspects of the figure members in the slope patterns with the standard ones, this process can recognize 3-D objects better than 92 percent of accuracy.
1
Some of the metrics are blocked by your
consent settings
Thai intonation analysis in harmonic-frequency domain
(1998-12-01)
Kongkachandra, Rachada
;
Pansang, Seri
;
Sripramong, Thanwa
;
Kimpan, Chom
The paper describes the Thai Intonation Analysis System (TIAS) for speaker-independent that achieves 90-95% word accuracy by applying two techniques, intonation flow analysis and voice identity calculation. The intonation flow analysis considers spectrograms emphasizing on their modification styles of the frequency peaks. The five styles of modification mapped to Thai tones as low, medium, falling, high, and rising respectively. However, there are some words that tones cannot be labeled correctly by the former because of speakers' amplitude, frequency, and uttering period differences. These limitations can be cured by applying the voice identity calculation that applies correlative coefficient relation to find voice identity of each compared spectrograms. The paper also reported experimental results for the task of intonation analysis with ten Thai numeral words uttered three-time repetitions by ten male speakers' readiness. In this paper, the speech waveforms are adapted from time-domain signals to harmonic-frequency domain.
2

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results