公开(公告)号：JPH0984002A

公开(公告)日：1997-03-28

申请号：JP22868095

申请日：1995-09-06

Applicant: IBM

Inventor： ECHIGO TOMIO ,  MAEDA JUNJI ,  KO MASAKUNI ,  IOKA MIKIHIRO

IPC: H04N11/04 ,  G06T9/00 ,  H03M7/30 ,  H04N1/41 ,  H04N19/00 ,  H04N19/102 ,  H04N19/17 ,  H04N19/196 ,  H04N19/20 ,  H04N19/21 ,  H04N19/36 ,  H04N19/423 ,  H04N19/50 ,  H04N19/94 ,  H04N19/96 ,  H04N7/24

Abstract: PROBLEM TO BE SOLVED: To provide an encoding method for reproducing an image of high picture quality by permitting a user to select picture quality and data quantity in multiple stages and adding data to compression data which can be decoded. SOLUTION: An original image is divided at every object area having prescribed correlation and the hierarchy structure of the object area is decided 62. The object area is plane-approximated 63 with more than one polygonal plane so that a luminance value and the errors of colors become within prescribed thresholds. A primary difference image is obtained 64 by obtaining a difference between the original image and the image which is plane- approximated. An nth order (n>=1) difference image is encoded, compressed 65 and the image which is plane-approximated and the encoded nth other difference image are stored 66. The (n+1)th order difference image is obtained 67 from a difference between the image obtained by expanding the compressed nth order difference image with the nth order difference image.

公开(公告)号：JPH05108730A

公开(公告)日：1993-04-30

申请号：JP24232991

申请日：1991-08-29

Applicant: IBM

Inventor： IOKA MIKIHIRO ,  KUROKAWA MASAHITO

IPC: G06F12/00 ,  G06F17/30

Abstract: PURPOSE: To improve the degree of freedom of retrieval by inputting movement intended by a user as a retrieving condition in retrieving an animation image. CONSTITUTION: A movement information data base 21 is constructed, based upon a motion vector string extracted from a moving image. A user inputs a vector string to be a retrieving condition by moving a device such as a mouse capable of changing a spatial position (step 23). Then matching is executed and a scene including the inputted movement is retrieved (step 25). Thereby it is unnecessary for the user to describe movement to be the retrieving condition by language or a conditional-expression.

公开(公告)号：JPH0984052A

公开(公告)日：1997-03-28

申请号：JP22539095

申请日：1995-09-01

Applicant: IBM

Inventor： ECHIGO TOMIO ,  MAEDA JUNJI ,  KO MASAKUNI ,  IOKA MIKIHIRO

IPC: H04N11/04 ,  G06T9/00 ,  H03M7/30 ,  H04N1/41 ,  H04N19/593 ,  H04N19/94

Abstract: PROBLEM TO BE SOLVED: To provide highly efficient data compression with simple conversion by dividing an image into plural picture element areas, obtaining the luminance values of the respective picture element areas and the parameter of a color, merging the plural picture element areas and generating a primary block candidate. SOLUTION: Digital picture data is inputted to an image input means 11 and is divided into the plural picture element areas by a parameter making means 12. The luminance values and the parameter of the color are obtained on the respective picture element areas. A generation/classification means 13 merges the plural picture element areas and generates the primary block candidate and it is classified to one of the plural patterns. A linear approximation conversion part 14 groups the primary block candidate where the luminance value of the picture element area and the parameter of the color can linearly be approximated/converted as a primary block among the primary block candidates belonging to the same classification and a conversion coefficient is expressed by the parameter. Then, a secondary block is grouped and a multidimensional block parameter is finally obtained.

公开(公告)号：JPH07203303A

公开(公告)日：1995-08-04

申请号：JP31685593

申请日：1993-12-16

Applicant: IBM

Inventor： IOKA MIKIHIRO

IPC: H04N5/268 ,  G06T1/00 ,  G06T7/20 ,  H04N5/14 ,  H04N19/00 ,  H04N19/105 ,  H04N19/136 ,  H04N19/142 ,  H04N19/172 ,  H04N19/186 ,  H04N19/196 ,  H04N19/423 ,  H04N19/85 ,  H04N19/87 ,  G06T7/00 ,  H04N7/24

Abstract: PURPOSE: To supply data for easily detecting the switching of the scene of a moving image. CONSTITUTION: Pixel data 50 of a moving image are inputted, and stored in a frame buffer 18 as pixel values 52. A statistical processor 14 calculates the mean value μ and standard deviation σ of a luminance value by using all the pixel data 50 of one frame. A conversion table processor 16 sets a pixel conversion table indicating the correspondence of the pixel values 52 with normalized output pixel values 54 by using the means value μ and the standard deviation σ, successively reads the pixel values 52, and outputs the output pixel values 54 in one frame to a detector 12 by using the pixel conversion table. The detector 12 calculates a difference between the previous output pixel value 54 and the present output pixel value 54, and outputs a scene change detection signal 56 at the time of judging that the number of picture elements beyond a threshold value Vth is more than a prescribed value and equivalent to a frame whose scene is switched.