Takashi Iwamoto

Face recognition system using local autocorrelations and multiscale integration

In this paper we investigate the performance of a technique for face recognition based on the computation of 25 local autocorrelation coefficients. We use a large database of 11,600 frontal facial images of 116 persons, organized in training and test sets, for evaluation. Autocorrelation coefficients are computationally inexpensive, inherently shift-invariant and quite robust against changes in facial expression. We focus on the difficult problem of recognizing a large number of known human faces while rejecting other, unknown faces which lie quite close in pattern space. A multiresolution system achieves a recognition rate of 95%, while falsely accepting only 1.5% of unknown faces. It operates at a speed of about one face per second. Without rejection of unknown faces, we obtain a peak recognition rate of 99.9%. The good performance indicates that local autocorrelation coefficients have a surprisingly high information content.

Doped InGaP grown by MOVPE on GaAs

Abstract Doping properties of InGaP grown by MOVPE on GaAs have been studied by changing not only dopant flux, but also growth temperature and PH 3 flow rate. Using DMZn and H 2 Se as dopant sources, p-type ( p = 5 × 10 17 -10 19 cm -3 ) and n-type ( n = 10 18 - 10 19 cm -3 ) InGaP epitaxial layers have been grown reproducibly. InGaP p-n diodes grown by MOVPE have been characterized by the measurements of capacitance-voltage, photoluminiscence and electroluminescence.

Direct synthesis and characterizations of fct-structured FePt nanoparticles using poly(N-vinyl-2-pyrrolidone) as a protecting agent.

FePt alloy in a bulk state is well known as a magnetic material. FePt nanoparticles, which are protected by poly(N-vinyl-2-pyrrolidone) (PVP) and have a face-centered tetragonal (fct) structure at a size of a few nanometers in diameter, have been directly synthesized by a polyol process in high-boiling point tetraethylene glycol used as a reducing reagent for the reduction of Fe(III) acetylacetonate and Pt(II) acetylacetonate. Their magnetic properties (coercivity and saturation magnetization) were dependent on the size and made progress as their diameters increased. The size in diameter was easily controlled by altering the content of PVP, the time for refluxing, and reaction temperature. FePt nanoparticles showed diameter-dependent coercivities at room temperature and they abruptly increased at over 4 nm in diameter. Ferromagnetic FePt nanoparticles with an fct structure were also synthesized at relatively low reaction temperature without refluxing. Likewise, as-synthesized FePt nanoparticles prepared by refluxing at 251 degrees C for 3 h displayed the fct structure and clearly indicated the ferromagnetism at room temperature. Reaction kinetics such as long refluxing time and slow temperature elevation rate were found to be important key factors to synthesize the ferromagnetic FePt nanoparticles although the reaction temperature was very critical as well.

Metalorganic vapor phase epitaxial growth of In1−xGaxP

In1−xGaxP layers have been grown by low pressure metalorganic vapor phase epitaxy using triethylindium (TEIn), triethylgallium (TEGa) and phosphine as source materials. The epitaxial layers of In0.49Ga0.51P lattice matched to GaAs substrate, obtained by controlling the TEIn and TEGA flows, have featureless surface morphology and exhibit efficient photoluminescence at room temperature. The peak energy and half width of photoluminescence spectra were studied in relation to the growth conditions.

AlGaAs growth using trimethyl and triethyl compound sources

Abstract Growth of GaAs, AlAs and AlGaAs using source materials of TMG, TEG, TMA and TEA has been studied in a low pressure MOVPE system. The different growth rates of the binary compounds for different sources indicate that the growth is characterized by the diffusion of non-decomposed and non-reacted metalorganics through the stagnant layer. An enhanced incorporation of Al into the solid compared to that of Ga has resulted upon mixing the metalorganics for the growth of AlGaAs. A discussion on the results in terms of the alkyl exchange interaction between metalorganics is presented.

Organometallic Vapor Phase Epitaxial Growth of In1-xGaxP (x ∼0.5) on GaAs

In1-xGaxP (x~0.5) layers have been grown on GaAs by low-pressure vapor phase epitaxy using triethylindium (TEI), triethylgallium (TEG) and phosphine. The observation of surface morphology and the measurements of peak energy and half-width of photoluminescence spectra indicate that high quality layers lattice-matched to GaAs substrates can be grown by controlling the substrate temperature and the TEI and TEG flows. The best sample has shown a room-temperature photoluminescence efficiency comparable to that of LPE layers of the same composition.

Correlation between Surface Morphology and Electrical Properties of GaAs Grown by Metalorganic Chemical Vapor Deposition

Surface morphology, including growth pyramids and rough surface, formed on MOCVD grown GaAs epitaxial layers has been investigated for different ratios of As to Ga atoms during growth. Such a morphology is correlated with the electrical property of the film. More specifically, rough surface textures together with p-type conductivity of the undoped sample result simultaneously for smaller As to Ga ratios. The effect of the As to Ga ratio on surface morphology and conductivity-type conversion is different for different orientations of substrates. This fact and the formation of the growth pyramids are explained on the basis of a local deficiency of As at the growth surface.

FUZZY INFORMATION FUSION IN A FACE RECOGNITION SYSTEM

We describe and evaluate information fusion by fuzzy integration in a robust, high performance face recognition system. The system uses fuzzy integrals to combine classifiers operating at different image resolutions. Recognition is carried out by distance classification of transformed vectors of local autocorrelation coefficients. The transformation is determined by linear discriminant analysis. A large database of 11,600 images of 116 persons is used to determine the system performance. After being trained to recognize 60 persons, it is tested on images of all persons in the database. Both training and test stages use 50 images of each person. Under two different training schemes, it achieves peak recognition rates of 98.4% and 97.9%, respectively, accepting only 1.6% and 2.4% of the unknown faces. This exceeds the performance of any of the individual classifiers by at least 10%. Moreover, it exceeds earlier results obtained by multiple resolution averaging on the same database by at least 1.0%.

MOCVD Growth of InGaP on GaP Substrates and Its Application to Visible LEDs

InGaP has been grown on GaP substrates by low pressure metalorganic chemical vapor deposition (MOCVD) using triethylindium, triethylgallium and phosphine. A buffer layer with step-graded alloy composition is inserted between the GaP substrate and the InGaP layer with a constant composition to reduce the influence of lattice mismatch. A trial fabrication of In0.34Ga0.66P light-emitting diodes(LEDs) has indicated that the diode efficiency is about one order of magnitude lower than commercially available GaAsP yellow LEDs. Further improvement in the efficiency is also discussed.

Optically Pumped Laser Operation of InGaAsP/InGaP Double Heterostructures Grown by Metalorganic Chemical Vapor Deposition

InGaAsP/InGaP double heterostructures have been grown on GaAs(100) substrates by low pressure metalorganic chemical vapor deposition. We have carried out a lasing experiment with a cleaved sample of ~200 µm width by optical pumping. At high excitation levels, we have observed a stimulated emission of a wavelength of 717 nm at 150 k. The threshold power density is estimated to be about 4.5×104 W/cm2 at this temperature. We also mention the room-temperature lasing operation of similar double heterostructure samples.