角江 崇

In a three-dimensional display by computer-generated holograms (CGHs), fast CGH calculation is required. The wavefront recording plane (WRP) method is able to reduce the calculation amount by placing WRPs in the vicinity of an object. In this paper, we optimized arrangement of WRPs to accelerate CGH generation.

Electroholography requires grayscale representation of reconstructed 3-D image to realize 3-D television. We propose gradation representation of reconstructed 3-D image from the binary hologram drawn in black and gray on bit-plane. Finally, we succeed to display color gradation 3-D image reconstructed by the proposed method.

The practical use of electroholography is limited by the complexity of the computer-generated hologram. We propose real-time electroholography using the multi-GPU cluster system with 13 GPUs (NVIDIA GeForce GTX TITAN X). Finally, we succeed to display reconstructed 3-D movie consisting of 95,949 object points at about 30 fps.

In this paper, we describe our latest results for holographic projection, including lensless zoomable holographic projection, the speckle reduction technique which is referred to as random phase-free method, and color zoomable holographic projection with the random phase-free method and color space conversion method in order to increase the image quality and accelerate the calculation time of holograms.

Computer Generated Holograms (CGH) are generated on computers; however, a great deal of computational power is required because the quality of the image is proportional to the number of point light sources of a 3D object. The Wavefront Recording Plane (WRP) method is an algorithm that enables reduction of the amount of calculations required. However, the WRP method also has a defect; it is not effective in the case of a 3D object with a deep structure. In this study, we propose two improved WRP methods: "Least Square Tilted WRP method" and "RANSAC Multi-Tilted WRP method." (C) 2015 Optical Society of America

This paper shows the method to calculate a computer-generated hologram (CGH) for real scenes under natural light using a commercial light field camera, and shows the results of color reconstruction of the synthesized CGHs. The CGH calculation using light field camera is performed by converting four-dimensional light field captured with a light field camera into a complex amplitude distribution, and the converted complex amplitude distribution is propagated so as to generate an interference pattern. In color reconstruction, we calculated three CGHs with red, green and blue wavelengths and superposed reconstructed red, blue and green images to obtain reconstructed color images. We verified that color three-dimensional images were reconstructed by numerical and optical reconstructions of the synthesized CGHs.

We report a high-speed parallel phase-shifting digital holography system using a special-purpose computer for image reconstruction. Parallel phase-shifting digital holography is a technique capable of single-shot phase-shifting interferometry. This technique records information of multiple phase-shifted holograms required for calculation of phase-shifting interferometry with a single shot by using space-division multiplexing. This technique needs image-reconstruction process for a huge amount of recorded holograms. In particular, it takes a long time to calculate light propagation based on fast Fourier transform in the process and to obtain a motion picture of a dynamically and fast moving object. Then we designed a special-purpose computer for accelerating the image-reconstruction process of parallel phase-shifting digital holography. We developed a special-purpose computer consisting of VC707 evaluation kit (Xilinx Inc.) which is a field programmable gate array board. We also recorded holograms consisting of 128 x 128 pixels at a frame rate of 180,000 frames per second by the constructed parallel phase-shifting digital holography system. By applying the developed computer to the recorded holograms, we confirmed that the designed computer can accelerate the calculation of image-reconstruction process of parallel phase-shifting digital holography similar to 50 times faster than a CPU.

Holographic projection that utilizes holography to record and reconstruct images to be projected is an attractive technique because it inherently requires no lenses, which could lead to the development of an ultra-small projector. In addition, holographic projection has unique functions that are difficult to perform on other projections: multi-projection that projects a multi-image on plural screens (which are also applicable on tilted screens) and projection on arbitrary surface screens; however, holographic projection has considerable problems: speckle noise, hologram calculation time and zoom without using a zoom lens. In this paper, we begin with basics of holographic projection (how to calculate holograms and reconstruct the projected images, and why we need a random phase), and then describe our latest results: the speckle reduction technique, which is referred to as random phase-free method; multi-projection, and real-time holographic projection with the random phase-free method. In addition, our holographic projection is capable of performing the zoom function without using a zoom lens, using a numerical method called scaled diffraction which can calculate diffraction at different sampling rates on a projected image and hologram.

We propose an aerial projection system for reconstruction of floating 3D images based on computer-generated image holograms (CGIHs). The proposed system consists of a spatial light modulator (SLM) and two concave mirrors. The SLM displays CGIHs, and a reconstructed 3D image is formed near the SLM plane. Because the two concave mirrors are set face to face, the floating image of the reconstructed 3D image can be projected at the hole of upper concave mirror. The system can reconstruct and project not only still 3D images but also 3D motion pictures by switching CGIHs.

This paper numerically investigates the feasibility of lensless zoomable holographic multiple projections to tilted planes. We have already developed lensless zoomable holographic single projection using scaled diffraction, which calculates diffraction between parallel planes with different sampling pitches. The structure of this zoomable holographic projection is very simple because it does not need a lens; however, it only projects a single image to a plane parallel to the hologram. The lensless zoomable holographic projection in this paper is capable of projecting multiple images onto tilted planes simultaneously. (C) 2014 Elsevier B.V. All rights reserved.

Ptychography is a promising phase retrieval technique for visible light, X-ray and electron beams. Conventional ptychography reconstructs the amplitude and phase of an object light from a set of the diffraction intensity patterns obtained by the X-Y moving of the probe light. The X-Y moving of the probe light requires two control parameters and accuracy of the locations. We propose ptychography by changing the area of the probe light using only one control parameter, instead of the X-Y moving of the probe light. The proposed method has faster convergence speed. In addition, we propose scaled ptychography using scaled diffraction calculation in order to magnify retrieved object lights clearly. (C) 2014 Elsevier B.V. All rights reserved.

We propose a real-time spatiotemporal division multiplexing electroholography utilizing the features of movies. The proposed method spatially divides a 3-D object into plural parts and periodically selects a divided part in each frame, thereby reconstructing a three-dimensional (3-D) movie of the original object. Computer-generated holograms of the selected part are calculated by a single graphics processing unit and sequentially displayed on a spatial light modulator. Visual continuity enables a reconstructed movie of the original 3-D object. The proposed method realized a real-time reconstructed movie of a 3-D object composed of 11,646 points at over 30 frames per second (fps). We also displayed a reconstructed movie of a 3-D object composed of 44,647 points at about 10 fps. (C)2014 Optical Society of America

We have implemented a computer-generated hologram (CGH) calculation on Greatly Reduced Array of Processor Element with Data Reduction (GRAPE-DR) processors. The cost of CGH calculation is enormous, but CGH calculation is well suited to parallel computation. The GRAPE-DR is a multicore processor that has 512 processor elements. The GRAPE-DR supports a double-precision floating-point operation and can perform CGH calculation with high accuracy. The calculation speed of the GRAPE-DR system is seven times faster than that of a personal computer with an Intel Core i7-950 processor. (C) 2014 Society of Photo-Optical Instrumentation Engineers (SPIE)

We report fast computation of computer-generated holograms (CGHs) using Xeon Phi coprocessors, which have massively x86-based processors on one chip, recently released by Intel. CGHs can generate arbitrary light wavefronts, and therefore, are promising technology for many applications: for example, three-dimensional displays, diffractive optical elements, and the generation of arbitrary beams. CGHs incur enormous computational cost. In this paper, we describe the implementations of several CGH generating algorithms on the Xeon Phi, and the comparisons in terms of the performance and the ease of programming between the Xeon Phi, a CPU and graphics processing unit (GPU). Program Summary Program title: Xeon-Phi-CGH Catalogue identifier: AETM_v1_0 Program summary URL: http://cpc.cs.qub.ac.uk/summaries/AETM_v1_0.html Program obtainable from: CPC Program Library, Queen's University, Belfast, N. Ireland Licensing provisions: Standard CPC licence, http://cpc.cs.qub.ac.uk/licence/licence.html No. of lines in distributed program, including test data, etc.: 26 539 No. of bytes in distributed program, including test data, etc.: 6 144 291 Distribution format: tar.gz Programming language: C, C++. Computer: Intel Xeon Phi coprocessor. Operating system: Linux. Has the code been vectorised or parallelized?: Yes. CPU and many cores in Xeon Phi coprocessor. RAM: 256M bytes Classification: 6.1, 6.5, 18. External routines: Intel MKL Nature of problem: We describe how to program fast computation of computer-generated holograms (CGHs) and diffraction calculation using Xeon Phi coprocessors, released by Intel. We describe the implementations of several CGH generating algorithms on the Xeon Phi, and the comparisons in terms of the performance and the ease of programming between the Xeon Phi, a CPU and graphics processing unit (GPU). Solution method: FFT-based diffraction calculations, computer-generated-hologram by direct integration. (C) 2014 Elsevier B.V. All rights reserved.

In a three-dimensional display by a computer-generated hologram (CGH), fast computation of CGH is required. In this paper, in order to accelerate CGH generation, the following two methods are used the first method is band-limited double-step Fresnel diffraction. Compared with convolution-based diffraction, such as an angular spectral method, the proposed method requires less computational time and memory. The second method is a wavefront recording plane (WRP) method which reduces the calculation amount by placing WRPs in the vicinity of an object. We succeeded in speeding up CGH calculation by combining both methods.

Computer based holography such as electro-holography and digital holography advances because the resolution of an imaging display and a sensing device such as a liquid crystal display (LCD) and a charge coupled device (CCD) becomes higher. The higher resolution of an imaging device requires a more powerful computing system. On the other hand, a solid state drive (SSD) develops as a storage device of computers. We studied the effectiveness of an SSD for a large-scale digital holography calculation. When the calculation data scale exceeds the main memory capacity, the SSD system showed better performance compared with a hard disk drive (HDD) system at computational speed and stability.

We propose a technique capable of obtaining spectral information and three-dimensional information of objects. This technique is based on digital holography and spectral estimation technique. In this technique, multiple lasers operating at different wavelengths, such as red, green, and blue, are used to record the complex amplitude of the object in digital holography, and multiple reconstructed monochrome images can be obtained by each wavelength. Spectral estimation technique is used to estimate the spectral reflectance distribution of the object from the multiple reconstructed monochrome images. Through the spectral estimation technique, the spectral reflectance distribution of object can be obtained. The effectiveness of the proposed method was confirmed by a numerical simulation and an experiment.

An A4-sized parallel phase-shifting digital holography system was constructed and experimentally demonstrated. Parallel phase-shifting digital holography is a technique being capable of instantaneous recording of three-dimensional (3-D) image of moving object. Aiming at a 3-D motion picture measurement of moving object system that can be carried on sites where the measurement system is needed, we improved a previous portable parallel phase-shifting digital holography system. The size and weight of the improved portable system are 300 mm x 210 mm x 160 mm and 6 kg, respectively. Also, we experimentally recorded a series of holograms of a moving object and reconstructed motion pictures, which show the same event but are focused at each depth, from the holograms. Thus, the 3-D motion picture recording and reconstruction capability of the constructed system were successfully demonstrated. A 3-D motion picture at 10 frames/s was obtained by the system.

A calculation reduction method for color digital holography (DH) and computer-generated holograms (CGHs) using color space conversion is reported. Color DH and color CGHs are generally calculated on RGB space. We calculate color DH and CGHs in other color spaces for accelerating the calculation (e. g., YCbCr color space). In YCbCr color space, a RGB image or RGB hologram is converted to the luminance component (Y), blue-difference chroma (Cb), and red-difference chroma (Cr) components. In terms of the human eye, although the negligible difference of the luminance component is well recognized, the difference of the other components is not. In this method, the luminance component is normal sampled and the chroma components are down-sampled. The down-sampling allows us to accelerate the calculation of the color DH and CGHs. We compute diffraction calculations from the components, and then we convert the diffracted results in YCbCr color space to RGB color space. The proposed method, which is possible to accelerate the calculations up to a factor of 3 in theory, accelerates the calculation over two times faster than the ones in RGB color space. (C) 2014 Society of Photo-Optical Instrumentation Engineers (SPIE)

We proposed a real-time time-division color electroholography using parallel calculation on multi graphics processing unit (GPU) system. Finally, we succeed to display a reconstructed color 3-D object of around 4000 points per color in real-time.

We try to display the real-time reconstructed 3-D image consisting of a large number of object points. We propose spatiotemporal division multiplexing electroholography using multi-GPU cluster system with InfiniBand network. Finally, the proposed method realized a real-time reconstructed movie of a 3-D object composed of44,647points.

We implemented an algorithm for enlarging the viewing-zone angle of holographic images on GPU with the aim of speeding up the calculation. We achieved results showing that calculations on the GPU were 15 times faster than on the CPU.

反復最適化計算でカラーホログラムのスペックルノイズを軽減するGS法を適用する場合の計算量を削減するために，RGB色空間からYCbCr色空間への色空間変換と色差成分のダウンサンプリングを利用したカラーホログラムの計算量低減手法を適用し，その結果について報告する．

インテグラルフォトグラフィを用いた広視域なホログラムの生成アルゴリズムの高速化を目指している．今回はホログラム生成に必要なIP画像の作成処理をGPUに実装し，またその前処理を最適化することで，CPUと比較して7倍高速化することに成功した．

We propose acceleration of color computer-generated holograms (CGHs) from three-dimensional (3D) scenes that are expressed as texture (RGB) and depth (D) images. These images are obtained by 3D graphics libraries and RGB-D cameras: for example, OpenGL and Kinect, respectively. We can regard them as two-dimensional (2D) cross-sectional images along the depth direction. The generation of CGHs from the 2D cross-sectional images requires multiple diffraction calculations. If we use convolution-based diffraction such as the angular spectrum method, the diffraction calculation takes a long time and requires large memory usage because the convolution diffraction calculation requires the expansion of the 2D cross-sectional images to avoid the wraparound noise. In this paper, we first describe the acceleration of the diffraction calculation using "Band-limited double-step Fresnel diffraction," which does not require the expansion. Next, we describe color CGH acceleration using color space conversion. In general, color CGHs are generated on RGB color space; however, we need to repeat the same calculation for each color component, so that the computational burden of the color CGH generation increases three-fold, compared with monochrome CGH generation. We can reduce the computational burden by using YCbCr color space because the 2D cross-sectional images on YCbCr color space can be down-sampled without the impairing of the image quality.

We designed and developed a special-purpose computer named FFT-HORN for high-speed imaging by digital holography. We demonstrated the developed FFT-HORN accelerated the computational time of reconstruction processing in digital holography.

We developed a 24-bit color volumetric display based on light emitting diode (LED). This volumetric display consists of 8 x 8 x 8 LEDs array and can represent 24-bit color three-dimensional images. The light intensity of the LEDs is gradated by the pulse width modulation method. We designed a control circuit of the volumetric display with the field programmable gate array. We succeeded in displaying 24-bit color three-dimensional moving image.

We developed "Loop architecture" which can perform different length FFT without changing fundamental circuit composition. In order to perform FFT effectively, we adopted Mixed-radix FFT algorithm with radix-2 and radix-4 butterfly calculation. Furthermore, we developed "Pipelined butterfly calculation system" and "5-bank memory architecture" in order to perform Mixed-radix FFT effectively. As a result, we succeeded to develop the versatile FFT processing system with small circuit area and low latency. In addition, we implemented our circuit to Spartan-6 FPGA board and developed sound analysis system via USB.

In high-speed 3D imaging by using parallel phase-shifting digital holography, there is a problem that the computational cost is very huge. Therefore, we have studied and developed special purpose computers for accelerating the calculation of parallel phase-shifting digital holography to solve the problem. However, the computational circuit for parallel phase-shifting interferometry and the special-purpose computer for light propagation have been independently designed. In this paper, we aim to connect them in order to realize high-precision and high-speed 3D imaging system by parallel phase-shifting digital holography.

We have constructed two portable parallel phase-shifting digital holography systems. One is an A3-sized parallel phase-shifting digital holography system and the other is an A4-sized parallel phase-shifting digital holography system. The size and weight of the A3-sized system are 450 mm (L) x 250 mm (W) x 200 mm (H) and 7kg, respectively. The system was constructed by use of ready-made optical components. To reduce the size and weight of the A3-sized system, the A4-sized system was designed and constructed. The size and weight of the A4-sized system are 300 mm (L) x 210 mm (W) x 160 mm (H) and 6 kg, respectively. The 3D motion picture recording and reconstruction capability of the constructed system were successfully demonstrated at 10 frames/second.

Scalar diffraction calculations, such as the angular spectrum method (ASM) and Fresnel diffraction, are widely used in the research fields of optics, x rays, electron beams, and ultrasonics. It is possible to accelerate the calculation using fast Fourier transform (FFT); unfortunately, acceleration of the calculation of nonuniform sampled planes is limited due to the property of the FFT that imposes uniform sampling. In addition, it gives rise to wasteful sampling data if we calculate a plane having locally low and high spatial frequencies. In this Letter, we developed nonuniform sampled ASM and Fresnel diffraction to improve the problem using the nonuniform FFT. c 2013 Optical Society of America

This erratum amends the lack of important works [P. Ferraro, Opt. Lett. 29, 854 (2004)] and [M. Paturzo, Opt. Express 18, 8806 (2010)] for scaled Fresnel diffraction in the reference. In order to magnify the reconstructed image from a hologram, Refs. [1, 2] use zero-padding and resampling of a hologram, respectively. (C) 2013 Optical Society of America

Projectors require a zoom function. This function is generally realized using a zoom lens module composed of many lenses and mechanical parts however, using a zoom lens module increases the system size and cost, and requires manual operation of the module. Holographic projection is an attractive technique because it inherently requires no lenses, reconstructs images with high contrast and reconstructs color images with one spatial light modulator. In this paper, we demonstrate a lensless zoomable holographic projection. Without using a zoom lens module, this holographic projection realizes the zoom function using a numerical method, called scaled Fresnel diffraction which can calculate diffraction at different sampling rates on a projected image and hologram. © 2013 Optical Society of America.

We demonstrate an in-line digital holographic microscopy using a consumer scanner. The consumer scanner can scan an image with 4,800 dpi. The pixel pitch is approximately 5.29 mu m. The system using a consumer scanner has a simple structure, compared with synthetic aperture digital holography using a camera mounted on a two-dimensional moving stage. In this demonstration, we captured an in-line hologram with 23, 602 3 18, 023 pixels (approximate to 0.43 gigapixels). The physical size of the scanned hologram is approximately 124 mm 3 95 mm. In addition, to accelerate the reconstruction time of the gigapixel hologram and decrease the amount of memory for the reconstruction, we applied the band-limited double-step Fresnel diffraction to the reconstruction.

We propose and demonstrate light-in-flight recording by parallel phase-shifting digital holography to obtain a motion picture of ultrashort light pulse propagation free from unwanted images, which are a zeroth-order diffraction image and a conjugate image, by phase-shifting interferometry with a single-shot exposure. We developed a polarization-imaging camera to record the propagation of a femtosecond light pulse whose center wavelength and temporal duration were 800 nm and 96 fs, respectively, and observed the propagation of the pulse without the unwanted images for 10 ps. (c) 2013 The Japan Society of Applied Physics

Numerical simulation of Fresnel diffraction with fast Fourier transform (FFT) is widely used in optics, especially computer holography. Fresnel diffraction with FFT cannot set different sampling rates between source and destination planes, while shifted-Fresnel diffraction can set different rates. However, an aliasing error may be incurred in shifted-Fresnel diffraction in a short propagation distance, and the aliasing conditions have not been investigated. In this paper, we investigate the aliasing conditions of shifted-Fresnel diffraction and improve its properties based on the conditions. © 2013 IOP Publishing Ltd.

Double-step Fresnel diffraction (DSF) is an efficient diffraction calculation in terms of the amount of usage memory and calculation time. This paper describes band-limited DSF, which will be useful for large computer-generated holograms (CGHs) and gigapixel digital holography, mitigating the aliasing noise of the DSF. As the application, we demonstrate a CGH generation with nearly 8K × 4K pixels from texture and depth maps of a three-dimensional scene captured by a depth camera. © 2013 Optical Society of America.

To improve the quality of reconstructed images, we apply bicubic interpolation and B-spline interpolation to parallel phase-shifting digital holography for the first time. The effectiveness of bilinear interpolation, bicubic interpolation, and B-spline interpolation in parallel phase-shifting digital holography is shown by a numerical simulation. In the simulation result, the application of bicubic interpolation and B-spline interpolation succeeded in decreasing the rootmean- square error of the reconstructed image by 12.6 and 11.9%, respectively.

Computer holography addresses holographic nature on a computer by computing light propagation. Its applications include digital holography, computer-generated holograms and so forth. Acceleration is an important issue because recent computer holography requires real-time processing or large-scale holograms, due to the progress of camera and display devices. In this paper, we present our acceleration techniques for computer holography using hardware (FPGAs, GPUs and Intel XeonPhi) and algorithms. © OSA 2013.

We succeeded in enlarging the viewing zone of a holographic image which was generated from an image captured by integral photography (IP). Although the quality of the reconstructed holographic image was somewhat degraded, we confirmed that we could observe the image of a large viewing zone (∼15 degrees) by using a lens array having a short focal length (2.2 mm).

Parallel phase-shifting digital holography (PPSDH) is a technique of single-shot phase-shifting digital holography. We found that there are two problems with this technique. (1) Some extraneous noises caused by the intensity unevenness of the reference wave become slightly superimposed on the object image. (2) The conjugate image cannot be completely removed. This is because the object wave causes the phase-shift error by illuminating an image sensor with a large incident angle. To solve these problems, we propose an algorithm for removing residual 0th-order diffraction and conjugate images in PPSDH. In the proposed algorithm, we modified phase-shifting interferometry in order to work through the unevenness of the intensity distribution and applied the Fourier transform technique to PPSDH to remove the residual conjugate image. The effectiveness of the proposed algorithm was experimentally verified. © 2013 The Optical Society of Japan.

We propose an image reconstruction algorithm for recovering high-frequency information in parallel phase-shifting digital holography. The proposed algorithm applies three kinds of interpolations and generates three different kinds of object waves. A Fourier transform is applied to each object wave, and the spatial-frequency domain is divided into 3 x 3 segments for each Fourier-transformed object wave. After that the segment in which interpolation error is the least among the segments having the same address of the segment in the spatial-frequency domain is extracted. The extracted segments are combined to generate an information-enhanced spatial-frequency spectrum of the object wave, and after that the formed spatial-frequency spectrum is inversely Fourier transformed. Then the high-frequency information of the reconstructed image is recovered. The effectiveness of the proposed algorithm was verified by a numerical simulation and an experiment. (C) 2012 Optical Society of America

We demonstrate motion pictures of femtosecond light pulse propagation. We adopted digital light-in-flight recording by holography as a technique for observation of femtosecond light pulse propagation. We recorded and reconstructed a moving picture of femtosecond light pulse propagating on a diffuser plate on which a test chart pattern was printed. The center wavelength and the duration of the light pulse were 800 nm and 96 fs, respectively. We successfully observed femtosecond light pulse propagation for 530 fs by the technique. © 2013 SPIE.

We present a special-purpose computer named HORN (HOlographic ReconstructioN) for fast calculation of computer- generated holograms (CGHs). The HORN can realize parallel processing of the CGH calculation by using field- programmable gate arrays. The latest version of HORNs, HORN-7, can reconstruct holographic images more clearly than previous HORNs because HORN-7 can make CGHs as a phase-only hologram (kinoform). In addition, the HORN- 7 can directly output calculated CGHs on a spatial-light modulator via Digital Visual Interface. In this paper, we demonstrate real-time reconstruction of holographic motion pictures by the HORN-7. We calculated CGHs, which consist of 1,920 × 1,080 pixels, from the object data of ∼6,000 points, and succeeded in reconstructing holographic motion pictures from the calculated CGHs at the rate of ∼7 frames per second. © 2013 Copyright SPIE.