The Journal of The Institute of Image Information and Television Engineers
Online ISSN : 1881-6908
Print ISSN : 1342-6907
ISSN-L : 1342-6907
Volume 64, Issue 3
Displaying 1-30 of 30 articles from this issue
Focus
Special Edition
The Latest Trend of Image Sensor Technologies
1. Review on 2009 IISW
2. 3D-real-time CCD Imager Based on Background-level Subtraction Scheme
3. Progress in Development of HARP Imaging Device and Its Application
Technical Survey
Technical Guide
Recent and Technologies in Multimedia Retrieval(3)
Keywords you should know(50)
My Recommendations on Research and Development Tools(32)
University of Ventures
Writing State-of-the-Art Image/ Video Processing Algorithms within 100 Lines(3)
News
  • Noriko Ide, Nana Akahane, Shigetoshi Sugawa
    2010 Volume 64 Issue 3 Pages 335-342
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    We discuss the operation methods for keeping a high S/N ratio at all switching points in multiple exposures and for shorting the total integration time of a wide dynamic range (DR) lateral overflow integration capacitor (LOFIC) CMOS image sensor that combines electric-charge voltage conversions in a pixel and a column capacitor. We have used these methods on a color LOFIC CMOS image sensor. This sensor had a pixel pitch of 5.6-um and the number of effective pixels was 800 (H) x 600 (V). It was made by using 0.18-um 2P3M CMOS technology. Fully linear responses were obtained, as well as a DR of 207 dB. Furthermore, the S/N ratio was 26-dB for the image of 18% gray card at all the switching points between the multiple exposures. The total exposure time was1/13-sec and was obtained by three photoelectric conversion operations at FD and LOFIC capacitors and one photoelectric conversion operation in the column capacitor.
    Download PDF (629K)
  • Hiroaki Fujita, Eric Stevens, Hirofumi Komori, Hung Doan, Jeffery Kyan ...
    2010 Volume 64 Issue 3 Pages 343-346
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    A new CMOS image sensor using a hole based detector has been developed. The measurement shows that crosstalk is reduced to one-third and dark current is reduced to a thirtieth that of a standard nMOS device. Red-into-green pixel crosstalk is 7% at 650 nm for a 4.3-μm pixel and dark current as low as 6 pA/cm2 at 60°C. Charge capacities of 60, 11 and 4kh+ have been measured for 4.3-, 1.75- and 1.4-μm pixels, respectively.
    Download PDF (814K)
  • Toshikazu Ohno, Mamoru Arimoto, Hayato Nakashima, Kaori Misawa, Yugo N ...
    2010 Volume 64 Issue 3 Pages 347-352
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    A charge multiplication CMOS image sensor was developed that has a 10μm pixel pitch and CIF format. This sensor has an electric charge multiplication electrode unit just behind the photodiode of each pixel, which repeatedly performs impact ionization against a photoelectrically-converted signal charge for the charge multiplication. A prototype sensor was used to evaluate the multiplication characteristics and it showed that the sensor can provide excellent signal amplification control in accordance with the repeated number of impact ionizations and the voltage applied to the charge multiplication electrode. This evaluation also showed that the spatial noise does not increase in relation to the signal amplification. In addition, the sensor yields a clearer image and a significantly higher S/N in a 0.4 lx low-light environment, in comparison with non-multiplying using charge multiplication of approx. 60 times in 100μs.
    Download PDF (685K)
  • Naoki Ozawa
    2010 Volume 64 Issue 3 Pages 353-357
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    A low noise driving method for CCD imaging devices is developed. In the method, a reset pulse is controlled according to the signal level of the CCD. By an evaluation employing a CCD of VGA size, it has become apparent that signal charge acquired from more than 100 pixels can be read continuously without reset pulses at a scene of 1/64 luminance and that low frequency noise whichi is conspicuous in a reproduced picture can be reduced remarkably.
    Download PDF (566K)
  • Kazuhiro Shimonomura, Tetsuya Yagi
    2010 Volume 64 Issue 3 Pages 358-364
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    We describe an intelligent vision sensor (silicon retina) that can pre-process images in real-time by using a parallel analog circuit that mimics the structure of the neuronal circuits in vertebrate retina. In this study, we have designed and fabricated a wide dynamic range silicon retina on the basis of our previous design. The photosensor of the present chip approximates logarithmic illumination-to-voltage transfer characteristics due to a photodiode capacitance modulation technique being applied. The chip in this study has achieved a dynamic range wide enough to perceive objects in both indoor and outdoor environments. Moreover, combining the logarithmic response of the photosensor with the Laplacian-Gaussian-like spatial filtering by a resistive network makes the response amplitude of the chip independent of the illumination intensity.
    Download PDF (2163K)
  • Hokuto Seo, Satoshi Aihara, Masakazu Namba, Toshihisa Watabe, Hiroshi ...
    2010 Volume 64 Issue 3 Pages 365-371
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    In order to achieve compact Super Hi-Vision color camera without the color separation optical systems, we have been developing a new type of image sensor with three organic photoconductive films sensitive to red(R), green(G) and blue(B). In this paper, we fabricated a stacked organic image sensor with G and R sensitive organic photoconductive films, each of which had readout circuit of zinc oxide (ZnO) thin film transistor (TFT). We successfully demonstrated an image pickup at TV flame rate. The sensor produced a color image with a resolution corresponding to the pixel number of ZnO TFT.
    Download PDF (2437K)
  • Tomonari Sawada, Kana Ito, Shoji Kawahito
    2010 Volume 64 Issue 3 Pages 373-380
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    The range resolution of a time-of-flight (TOF) range image sensor with periodical charge-draining operation is analyzed under the influence of ambient light. The analysis explains the effectiveness of a small duty light pulse and a periodical charge-draining structure to reduce the influence of ambient light. A theoretical model of range resolution taking into consideration factors of the duty ratio, the ambient light component, and the diffusion component is presented. The theoretical model under the influence of ambient light well agrees with the experimental results. The offset-charge-canceling technique using two sub-frames effectively enables accurate range measurements under ambient light.
    Download PDF (2359K)
  • Takenori Hara, Hideo Saito, Takeo Kanade
    2010 Volume 64 Issue 3 Pages 381-388
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    View disturbing noise must be removed from images obtained with a camera under adverse weather conditions. In this paper, we present a method for removing glare caused by water droplets, or other foreign objects, adhering to protective glass of an imaging lens. We have designed and implemented an electronically controlled optical shutter array that detects and removes glare. We also present the possibility of applying this technique to remove general glare caused by the imaging lens itself.
    Download PDF (2731K)
  • Masaki Igarashi, Masayuki Ikebe, Sousuke Shimoyama, Kenta Yamano, Juni ...
    2010 Volume 64 Issue 3 Pages 389-398
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    We propose a fast 2-D bilateral filtering method without kernel-size dependency. Focusing on weighted local histograms and the central limit theorem combined with applying line buffers of column histograms, enables us to reduce the number of necessary memory accesses and calculations. Numerical experiments demonstrated a reduction in both calculations and kernel-size independency. We used a dual core 2-GHz CPU with our method and were able to achieve one million pixels per 0.5 sec operation without the need for downsampling, Single Instruction/Multiple Data (SIMD) or multi-thread operation.
    Download PDF (3019K)
  • Yoshiro Takiguchi, Masakazu Nanba, Yuuki Honda, Misao Kubota, Norifumi ...
    2010 Volume 64 Issue 3 Pages 399-404
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    We have been developing an image sensor, "FEA-HARP", which consists of a field emitter array (FEA) and high-gain avalanche rushing amorphous photoconductor (HARP). This sensor will be used in an ultrahigh-sensitivity compact HDTV camera. We propose new operation methods for the FEA-HARP that maintain picture quality even when strong light enters the sensor. Experimental results show that blooming generation was suppressed by removing excess charges from the HARP target during horizontal blanking intervals.
    Download PDF (1848K)
  • Hiroshi Ikeoka, Takayuki Hamamoto
    2010 Volume 64 Issue 3 Pages 405-412
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    We have been investigating real-time depth estimation techniques of a wide detectable range. These techniques involve the use of tilted optics imaging to use variance of depth of field on the horizontal planes of sharp focus. They need far fewer multiple focus images than the conventional passive methods do, e.g., the Depthfrom-Focus and the Depth-from-Defocus. Hence, our method gets rid of the bottleneck of conventional methods; the motion speed of optical mechanics is significantly slower than that of image processing parts. Therefore, it is suitable for applications that involve estimation of depth and that need a wide detectable range and real-time processing such as automation tasks.
    Download PDF (783K)
  • Junichi Akita, Yui Maeda
    2010 Volume 64 Issue 3 Pages 413-415
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    Pixels in conventional imaging systems are located in lattice sites, and this lattice placement of pixels causes jaggy artifacts in the image, especially at the edges of slanted lines with high-intensity contrast, which often results in severe defects in the image quality. The conventional approaches to overcoming this problem of jaggedness are to increase the number of pixels and to use anti-aliasing. However, the number of pixels is limited by the physical pixel size and the quantity of image information, while anti-aliasing intrinsically causes blurred images. The authors have been researching and evaluating image systems with pseudorandom pixel placement to reduce the effects of jaggy artifacts. We describe here the design of an image sensor that contains pixels with both pseudorandom and convectional lattice placement using an identical active pixel sensor (APS) pixel circuit. We also describe the preliminary experimental results obtained by testing the fabricated image sensors.
    Download PDF (2474K)
  • Takayuki Kimura, Yutaro Yokoyama, Toru Masuzawa
    2010 Volume 64 Issue 3 Pages 416-418
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    To improve the frame rate of a two-dimensional integrated magnetic sensor, the pixel structure and readout circuits were re-examined. The frame rate was improved by reducing the noise of the horizontal shift register by lowering the drive frequency and parallel reading (with 16 channels). With this idea, two-dimensional integrated magnetic sensors were designed and fabricated with the standard 0.35 μm CMOS process on silicon. The type of Hall sensor is n-type Hall sensor that uses an inversion layer under the gate oxide of the MOSFET. The Hall sensors were arrayed (64×64), and the control digital circuits and output amplifier were also integrated into the same chip. “One pixel” was 50×50 μm, and the entire chip was 4.9×4.9 mm. The frame rate was 7813 frames/sec at a 2-MHz horizontal shift register frequency. The average sensitivity of these sensors was 72.9 mV/(mA·T). The two-dimensional magnetic flux distribution was measured with a 3000-rpm, 1-mm-diameter Nd-Fe-B rare-earth permanent rotating magnet without image lag. From the measurement results, a high frame-rate magnetic sensor for motor control was successfully fabicated.
    Download PDF (1698K)
  • Yunkyung Kim, Makoto Ikeda, Kunihiro Asada
    2010 Volume 64 Issue 3 Pages 419-422
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    With the downscaling of CMOS technology, interconnect layers are multi-stratified since the number of metal levels has increased. However, this multi-metal-layer structure above the photodiodes affects sensitivity, which is important for obtaining optimal performance from CMOS image sensors. This paper analyzes the light-transmission characteristics on multi-metal layers (interconncet layers). To evaluate what effect standard CMOS-process technologies have on multi-metal layers, we developed a method of calculating the transmitted light intensity through the multi-metal layers. We found the transmitted light intensity from the calculated results using the transfer matrix method for standard CMOSs ranging in size from 1.2 μm to 22 nm.
    Download PDF (1044K)
  • Yongqing Sun, Satoshi Shimada, Masashi Morimoto, Yukinobu Taniguchi
    2010 Volume 64 Issue 3 Pages 423-434
    Published: March 01, 2010
    Released on J-STAGE: June 01, 2010
    JOURNAL FREE ACCESS
    In this paper we present a novel approach to modeling visual concepts effectively and automatically using web images. The selection of training data (positive and negative samples) is strongly related to the quality of learning algorithms and is an especially crucial step when using noisy web images. In this scheme, first, images are represented by regions from which training samples are selected. Second, region features effectively representing a semantic concept are determined, and on their basis, the representative regions corresponding to the concept are selected as reliable positive samples. Third, high quality negative samples are determined using the selected positive samples. Last, the visual model associated with a semantic concept is built through an unsupervised learning process. The presented scheme is completely automatic and performs well for generic images because of its robustness in learning from diverse web images. Experimental results demonstrate its effectiveness.
    Download PDF (5077K)
feedback
Top