All issues
Predecessor
21.42
Displaying 1-16 of 16 articles from this issue
- |<
- <
- 1
- >
- >|
-
Article type: Cover
Pages Cover1-
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (16K) -
Article type: Index
Pages Toc1-
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (64K) -
Junji ONISHI, Kineo MATSUI, Shinji OZAWAArticle type: Article
Pages 1-5
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this report, we propose how to detect a signature from clipped picture in which is already embedded as watermark. The part of Multi-resolution apporximate (MRA) is used in order to combine the original signed picture with the clipped picture. The signature is extracted from those combined pictures. The wavelet transform is a good tool that makes a clipped part fit to the original one smoothly. Our scheme provides hard copy right protection for distibuting clipped pictures.View full abstractDownload PDF (620K) -
Motonori Doi, Akio Ueda, Kosuke Sato, Kunihiro ChiharaArticle type: Article
Pages 7-12
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThe prevention of illegal intruding into buildings is very important. This paper presents two security systems for the prevention of illegal intruding using image processing. One is an intruder detection system using infrared camera and range finder, and another is a lock-control system using face identification. In the intruder detection system, first, the intruder is judged by comparing his silhouette with human features dictionary, and measuring the height of intruder using a range data. Secondly, the intruder is tracked by its motion vector, size and average intensity. The lock-control system identify a face features in normalized face image for judgment whether it should unlock a door or not. Experiments show that these system is suitable for automated security systems.View full abstractDownload PDF (728K) -
Tomoko MATSUIArticle type: Article
Pages 13-18
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThis paper describes the basic structure of a speaker verification system and introduces technical issues on updating a reference template/model and on normalizing similarity values. To update a reference template/model, a method is described that time-registers the latest utterances and averages them. Another method is described for updating a hidden Markov model (HMM) for each speaker using a small amount of new data for each speaker so that they are close to the parameter values given by the maximum likelihood estimation using all data. For normalizing similarity values, methods based on the likelihood ratio and the a posteriori probility, and a method using the weighting models rank are introduced. Each method reduces the recognition error rates to less than half under each experimental condition using data including utteruce variations.View full abstractDownload PDF (728K) -
Shunsuke Takata, Hiroyuki Matsumoto, Syuji Kawakubo, Kiyoshi YamanakaArticle type: Article
Pages 19-24
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe have developed an LSI, which enables multi encryption algorithm. Our LSI is a single-chip computer consisting of RISC-CPU, Flash-ROM, RAM and so on. It supports secret-key cryptosystems, public-key cryptosystems and message digest functions. Our chip is constructed so as session-key of secret-key cryptosystems or private-key of public-key cryptosystems themselves do not reveal, while encryption, authentication or signature algorithms are performed. Also, it has a function or detecting altered programs in Flush-ROM. In this paper, we describe mainly those characteristics of our LSI and give an examples of image data distribution system as one of applications.View full abstractDownload PDF (686K) -
Aboul Ella Hassanien, Hiroki TAKAHASHI, Masayuki NAKAJIMAArticle type: Article
Pages 25-30
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSImage metamorphosis (For short, image morphing) is an active field of research and recent efforts aim at improving both user interface and warping results. Generally, the morphing technique involves three problems. The first is to establish correspondence between given two images, which is most difficult part of morphing process. The correspondence is usually established by hand. The second problem is to define or construct a mapping function which deforms the first image towards the second one based on these feature points, and vice versa. The third problem is to blend pixel values of the two deformed images to create in-between images. this will end the morphing process. This study aims to raise strategy to solve these problems. First, we propose a semi-automatic algorithm based on snake model to specify the feature correspondence between two given images. It allows a user to extract contour that define facial feature such as lips, mouth, profile. etc., by specifying only endpoints of the contour around the feature that serves as the extremities of contour. Then we use these two points as anchor points, and automatically computes the image information around these endpoints to provide boundary conditions. Next, we optimize the contour by taking this information into account close to its extremities. During the iterative optimization process, the image forces are moving progressively from the contours extremities towards its center to define the feature. It helps the user to define easily the exact position of a feature. It may also reduce the time taken to establish feature correspondence between two images. For the second image morphing problem, this paper presents a new image warping algorithm which uses elastic body spline to generate a warp function that interpolates scattered data points. The spline is based on a Partial Differential Equation of Naiver that describes the equilibrium displacement of the elastic body objected to forces. It generates a smooth warp that reflects the feature correspondence points and has several advantages. It is efficient in smoothed interpolated morphed images with only a remarkably small number of feature points specified. Finally, we describe an efficient cross dissolve function to generate the in-between images. The in-between images could be defined by creating a new set of feature points through the cross-dissolving process.View full abstractDownload PDF (858K) -
[in Japanese], [in Japanese], [in Japanese], [in Japanese], [in Japane ...Article type: Article
Pages 31-36
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSPhase Only Correlation is a pattern matching algorithm which replaces magnitude components of two input images with a constant in the process of calculating ordinary correlation. POC is known to have good discrimination characteristics producing intense peak on correlation domain if two images have common components. For images in which phase components are dominant, manipulation in correlation domain such as inclusion or exclusion of a peak have an filtering effect on spatial images, such as enhancement of spatial signals or noises. New filtering algorithm named Correlation Domain Filtering is proposed.View full abstractDownload PDF (784K) -
Masahide NAEMURA, Yasunobu MIZUTANI, Atsushi FUKUDA, Yoshinori IZUMIArticle type: Article
Pages 37-42
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe propose a new method to detect lawn areas from sport programs (soccer). The main feature of yhe method lies in that the detection is realized by two-stage operations that consists of creation of color histograms and geodesic binary reconstruction. The color histograms is manually created with prior knowledge about color information in reheasal time. Geodesic binary reconstruction is one of morphological operators and can realize a stable automatic detection, using two different types of lawn areas that are created by two threshold operations of the color histograms.View full abstractDownload PDF (1165K) -
Masafumi SATO, Hirotsugu KINOSHITAArticle type: Article
Pages 43-48
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper, we propose a new digital signature system for images. This system generates the signature information based on the structured information of the image. It increases the tolerance resistance against processing and coding work, and the coding. An image possesses a structured information both physically and semantically. This characteristics is deployed as an image retrievial method which allows rough sketch matching. This system has compatibility with retrieval methods which are based on the structured information.View full abstractDownload PDF (740K) -
Hiroshi SHIMAMOTO, Masayuki SUGAWARA, Kohji MITANI, Yoshihiro FUJITAArticle type: Article
Pages 49-54
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSAbstruct-An enhancer for very-high resolution acuisition system with quinquanx sampling method is studied. We simulated on the computer with image data which is captured from the live pictures by using our developed very-high resolution camera. Five taps two-dimensional FIR filter enhanced 1500TV lines or more without moire. Also, our seven taps interpolation eliminates moires in the diagonal direction but keeps high-resolution in the horizontal and vertical directions, and is suitable for this quinquanx sampling system.View full abstractDownload PDF (994K) -
Tetsushi Tobita, Hiroshi Tadokoro, Shin-ichi MurakamiArticle type: Article
Pages 55-60
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSA facial expression is defined by a time sequence of facial images. Then, the various kinds of facial expressions can be generated from a basic facial image sequence by modifying the time generation functions. This paper describes a 3D image generation method for facial expressions by use of 3 kinds of time generation function. The generated facial expressions by these functions are compared. And a decomposition method of observed facial expressions into these time generation functions are also presented.View full abstractDownload PDF (1098K) -
Masanori Yamada, Kazuyuki Ebihara, Jun OhyaArticle type: Article
Pages 61-66
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn our "Virtual KABUKI System", to estimate human body postures, a human silhouette must be segmented from a background image. We had used an infrared camera to segment it. But clothes which are hard to have a heat, such as a skirt and a jacket, obstruct a detection of a complete human silhouette. To solve this problem, we developed the new method which uses a difference between a current image and a background image. This paper introduce a new real-time method to estimate human body postures from color images acquired by a normal CCD camera.View full abstractDownload PDF (1286K) -
Katusyuki ShinoharaArticle type: Article
Pages 67-72
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this thesis, the done problem was arranged concerning "Individual identification method by the pinna image which used morphological operation" which had been proposed before.Moreover, the morphology individual with the pinna investigated to identify the individual based on the characteristic of the segment and the area where the element of the pinna was represented and the result was shown with coefficient of variation concerning the pinna element.View full abstractDownload PDF (939K) -
Isao IshiwataArticle type: Article
Pages 73-75
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThe retinal vascular pattern of human eye is a unique characteristic and remained stable for an individual life. This vascular pattern can be used for personal recognition. This paper explains how to get the 96 bite individual tempplate to record and use it for high accuracy (1ppm) personal identification.View full abstractDownload PDF (445K) -
Article type: Appendix
Pages App1-
Published: July 22, 1997
Released on J-STAGE: June 23, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (70K)
- |<
- <
- 1
- >
- >|