公开(公告)号：KR20070094521A

公开(公告)日：2007-09-20

申请号：KR20070025700

申请日：2007-03-15

Applicant: SONY CORP

Inventor： KONDO TETSUJIRO ,  MIYAI TAKESHI ,  UCHIDA MASASHI ,  MORIMURA TAKUO ,  CHIKAOKA SHIZUO ,  KIKUCHI DAISUKE ,  KUNIHIRO TAKESHI ,  MORI HIDEKI

IPC: H04N7/01

CPC classification number: H04N7/012 ,  H04N5/145 ,  H04N7/014 ,  H04N7/0145

Abstract: An image processing apparatus, an image processing method and a program are provided to convert an interlaced image into a progressive image with high picture quality. An image processing apparatus includes a converter(121), a motion vector detector(131), a cyclic coefficient setting unit(133), a motion compensator(134), and an output image generator. The converter converts an input interlaced image into a progressive intermediate image. The motion vector detector detects a motion vector of the input image which has a distance shorter than a pixel interval of the intermediate image as a minimum unit. The cyclic coefficient setting unit sets a first cyclic coefficient for first type pixels located in a portion of the input image where pixels exist and a second cyclic coefficient for second type pixels located in a portion of the input image where pixels do not exist. The motion compensator motion-compensates a previous output image to generate a motion-compensated image. The output image generator adds up values of the first type pixels and values of pixels corresponding to the motion-compensated image and adds up values of the second type pixels and the values of the pixels corresponding to the motion-compensated image to generate an output image.

Abstract translation: 提供图像处理装置，图像处理方法和程序以将隔行扫描图像转换成具有高图像质量的逐行图像。 图像处理装置包括转换器（121），运动矢量检测器（131），循环系数设定单元（133），运动补偿器（134）和输出图像生成器。 转换器将输入隔行扫描图像转换为逐行中间图像。 运动矢量检测器检测出距离比中间图像的像素间隔短的输入图像的运动矢量为最小单位。 循环系数设定单元设置位于输入图像的存在像素的部分的第一类型像素的第一循环系数和位于不存在像素的输入图像的一部分中的第二类型像素的第二循环系数。 运动补偿器运动补偿先前的输出图像以产生运动补偿图像。 输出图像生成器将与运动补偿图像相对应的第一类型像素和像素值相加，并将第二类型像素的值和与运动补偿图像相对应的像素的值相加以生成输出图像 。

公开(公告)号：KR20070094566A

公开(公告)日：2007-09-20

申请号：KR20070026226

申请日：2007-03-16

Applicant: SONY CORP

Inventor： KONDO TETSUJIRO ,  UCHIDA MASASHI ,  SUGA YASUHIKO ,  HOSOKAWA KENICHIRO

IPC: G06T5/00 ,  G06T3/00 ,  G06T5/50 ,  G06T7/40 ,  G06T7/60 ,  G06T9/00 ,  H04N1/40 ,  H04N7/015 ,  H04N7/24

CPC classification number: G06K9/00456 ,  G06K9/00624 ,  G06K9/4638

Abstract: An image processing apparatus and method, a program storage medium, and a program are provided to process a region including components of a natural image and a region including components of an artificial image respectively to improve the picture quality of each region. An image processing apparatus includes a broad-range feature quantity extractor(911), a broad-range artificiality calculator(912), a narrow-range feature quantity extractor(914), a narrow-range artificiality calculator(915), and an artificiality calculator(913). The broad-range feature quantity extractor extracts a broad-range feature quantity for a noted pixel of a first image. The broad-range artificiality calculator calculates degrees of artificiality according to position relation of the extracted broad-range feature quantity with respect to a statistical distribution range of an artificial image of the first image. The narrow-range feature quantity extractor extracts a narrow-range feature quantity from pixels located in a predetermined region for the noted pixel of the first image. The narrow-range artificiality calculator calculates degrees of artificiality according to position relation of the extracted narrow-range feature quantity with respect to the statistical distribution range of the artificial image. The artificiality calculator combines the broad-range degrees of artificiality and the narrow-range degrees of artificiality to generate degree of artificiality of the noted pixel.

Abstract translation: 提供图像处理装置和方法，程序存储介质和程序以分别处理包括自然图像的分量的区域和包括人造图像的分量的区域，以提高每个区域的图像质量。 一种图像处理装置包括宽范围特征量提取器（911），宽范围人造计算器（912），窄范围特征量提取器（914），窄范围人造计算器（915） 计算器（913）。 宽范围特征量提取器提取第一图像的标记像素的宽范围特征量。 宽范围人造计算器根据所提取的宽范围特征量相对于第一图像的人造图像的统计分布范围的位置关系来计算人造度。 窄范围特征量提取器从位于第一图像的所述像素的预定区域中的像素提取窄范围特征量。 窄范围人造计算器根据提取的窄范围特征量相对于人造图像的统计分布范围的位置关系来计算人造度。 人造计算器将广泛程度的人造和窄范围的人造学结合起来，以产生所述像素的人造度。

公开(公告)号：KR20070094520A

公开(公告)日：2007-09-20

申请号：KR20070025645

申请日：2007-03-15

Applicant: SONY CORP

Inventor： KONDO TETSUJIRO ,  UCHIDA MASASHI ,  MORIMURA TAKUO ,  KIKUCHI DAISUKE ,  MIYAI TAKESHI ,  KASAMA HIDEO ,  KUNIHIRO TAKESHI ,  NAKAMURA YOSHIAKI ,  MORI HIDEKI ,  SUGA YASUHIKO ,  HOSOKAWA KENICHIRO ,  CHIKAOKA SHIZUO

IPC: G06F3/12

CPC classification number: H04N7/014 ,  H04N5/144 ,  H04N7/012

Abstract: An image processing apparatus, an image processing method and a program thereof are provided to allow the accurate conversion of an input image into a high-quality image having the number of pixels different from that of the input image. An image processing apparatus comprises the followings. A converting unit(101), which includes an interlace-progressive conversion unit(111), a motion-vector detection unit, a cyclic-coefficient setting unit, a motion compensation unit and an output image generating unit, converts an interlace image composed of a first number of pixels into a first progressive image. An interpolation unit interpolates the first progressive image to generate a second progressive image composed of a second number of pixels and a number of pixels of a target image, and a classifying unit classifies subject pixels forming a third progressive image as the target image into classes in accordance with a feature of the second progressive image composed of the second number of pixels. A storing unit stores a prediction coefficient for each of the classes, and a computing unit determines the third progressive image as the target image from the second progressive image.

Abstract translation: 提供图像处理装置，图像处理方法和程序以允许将输入图像精确地转换成具有与输入图像的像素数不同的像素数量的高质量图像。 一种图像处理装置包括以下。 A转换单元（101），其包括交错逐行转换单元（111），运动矢量检测单元，循环系数设置单元，运动补偿单元和输出图像生成单元，将由 将第一数量的像素转换成第一渐进图像。 内插单元内插第一逐行图像以生成由第二数量的像素和目标图像的像素数构成的第二逐行图像，并且分类单元将形成作为目标图像的第三渐进图像的被摄体像素分类为 根据由第二数量的像素组成的第二逐行图像的特征。 存储单元存储每个等级的预测系数，并且计算单元从第二逐行图像确定第三逐行图像作为目标图像。

公开(公告)号：EP0843475A4

公开(公告)日：2000-04-12

申请号：EP97924276

申请日：1997-05-30

Applicant: SONY CORP

Inventor： HORISHI TAKASHI ,  UCHIDA MASASHI ,  KONDO TETSUJIRO

IPC: H04N7/01 ,  H04N7/26 ,  H04N7/46 ,  H04N7/50

CPC classification number: H04N7/0135 ,  H04N7/01 ,  H04N7/0125 ,  H04N7/0145 ,  H04N19/117 ,  H04N19/136 ,  H04N19/172 ,  H04N19/587 ,  H04N19/59 ,  H04N19/61 ,  H04N19/98

Abstract: An ADRC circuit (3) generates a space class from SD data segmented by an area segmenting circuit (2) and a motion class deciding circuit (5) generates a motion class from the SD data segmented by an area segmenting circuit (4). A class code generating circuit (6) generates a class code from the space class and motion class. The additional code data which degenerate the tap of the SD data at every class code are supplied to a tap degenerating circuit (10) from a tap degeneration ROM (7) and the circuit (10) degenerates the SD data segmented by an area segmenting circuit (9). An estimating arithmetic circuit (11) obtains HD data by estimating the linearity between the coefficient data corresponding to the class codes and supplied from a ROM table (8) and the degenerated SD data.

公开(公告)号：EP1282236A4

公开(公告)日：2009-07-08

申请号：EP01930004

申请日：2001-05-09

Applicant: SONY CORP

Inventor： KONDO TETSUJIRO ,  HAMAMATSU TOSHIHIKO ,  NAKAYA HIDEO ,  NISHIKATA TAKEHARU ,  OHTSUKA HIDEKI ,  KUNIHIRO TAKESHI ,  MORIFUJI TAKAFUMI ,  UCHIDA MASASHI

IPC: H03M7/30 ,  H03M7/36 ,  H04N1/41 ,  H04N7/26 ,  H04N7/30 ,  H04N7/32 ,  H04N7/50

CPC classification number: H03M7/30 ,  H04N19/117 ,  H04N19/14 ,  H04N19/176 ,  H04N19/44 ,  H04N19/46 ,  H04N19/60 ,  H04N19/61 ,  H04N19/63 ,  H04N19/86 ,  H04N19/98

Abstract: JPEG-coded data is entropy-decoded to quantized DCT coefficients, which are supplied to a prediction tap extracting circuit (41) and a class tap extracting circuit (42). the prediction tap extracting circuit (41) and the class tap extracting circuit (42) extract necessary ones from the quantized DCT coefficients to produce a prediction tap and a class tap. A class sorting circuit (43) conducts class sorting according to the class tap. A coefficient table storage unit (44) supplies the tap coefficient corresponding to the class determined by the class sorting to a product-sum operating circuit (45). The product-sum operating circuit (45) conducts a linear prediction operation by using the tap coefficient and the class tap to produce decoded image data.

公开(公告)号：EP1104190A4

公开(公告)日：2006-04-19

申请号：EP00919154

申请日：2000-04-21

Applicant: SONY CORP

Inventor： KONDO TETSUJIRO ,  TATEHIRA YASUSHI ,  UCHIDA MASASHI ,  HATTORI MASAAKI ,  MIYAI TAKESHI

IPC: H04N7/01

CPC classification number: H04N7/0135 ,  H04N7/0145

Abstract: An interlaced input image signal is supplied at a field frequency of 50 Hz. Class detection circuits detect classes corresponding to the patterns of the level distribution of input pixels near an output pixel to be produced. Prediction factor sets corresponding to the classes are read out of prediction factor memories. Product-sum operation circuits calculate data on the output image signals by using a linear estimation formula of estimation taps (pixels of the input image signals) and the prediction factor sets. The pixel values (M and S) of the output image signals of 50 Hz are outputted by the product-sum circuits. The pixel values (M and S) from the product-sum circuits are converted into signals of a frequency of 60 Hz. A selector selects one of the outputs of the field memories alternately and generates an output image signal (field frequency of 60 Hz).

公开(公告)号：EP0635978A4

公开(公告)日：1995-04-19

申请号：EP94902095

申请日：1993-12-09

Applicant: SONY CORP

Inventor： KONDO TETSUJIRO SONY CORPORATI ,  UCHIDA MASASHI ,  KAWAGUCHI KUNIO SONY CORPORATI

IPC: H04N7/01 ,  H04N7/12 ,  H04N7/26

CPC classification number: H04N7/0135 ,  H04N7/0125 ,  H04N7/125 ,  H04N19/98

Abstract: A digitally inputted image signal (SD signal) is converted into a high resolution digital image signal (HD signal) by using estimated values. The specific pixels of an object to be estimated are classified according to the one-, two-, or three-dimension level distribution of plural reference pixels of SD signal located near the specific pixels. The estimated values of the specific pixels are generated by a linear combination of the values of the reference values and the estimation coefficients determined in advance by learning. During the learning, a known HD signal and the SD signal formed from it are used to determine the estimation coefficients so as to minimize the sum of squares of the errors between the true values and the values estimated by the linear combination of the values of the surrounding SD signal reference pixels and the estimated coefficients. Not necessarily limited to the estimated coefficients, it may be possible to use, as the estimated values, representatives determined for every class in correspondence with the class of the inputted SD signal. The reference value of a block and the value normalized by the dynamic range DR are an example as the representatives.

Abstract translation: 数字输入的图像信号（SD信号）通过使用估计值被转换成高分辨率数字图像信号（HD信号）。 根据位于特定像素附近的SD信号的多个参考像素的一维，二维或三维电平分布来对待估计对象的特定像素进行分类。 通过学习预先确定的参考值和估计系数的值的线性组合来生成特定像素的估计值。 在学习期间，已知的HD信号和由其形成的SD信号被用于确定估计系数，以便最小化真值与通过线性组合值估计的值之间的误差的平方和 围绕SD信号参考像素和估计的系数。 不一定限于所估计的系数，可以使用对应于所输入的SD信号的类别为每个类别确定的代表作为估计值。 块的参考值和由动态范围DR归一化的值是作为代表的示例。

公开(公告)号：DE69731872D1

公开(公告)日：2005-01-13

申请号：DE69731872

申请日：1997-05-30

Applicant: SONY CORP

Inventor： HORISHI TAKASHI ,  UCHIDA MASASHI ,  KONDO TETSUJIRO

IPC: H04N7/01 ,  H04N7/26 ,  H04N7/46 ,  H04N7/50

Abstract: An ADRC circuit 3 generates spatial classes with SD data extracted by an area extracting circuit 2. A moving class determining circuit 5 generates a moving class with SD data extracted by an area extracting circuit 4. A class code generating circuit 6 generates a class code with the spatial class and the moving class. A tap decreasing ROM 7 supplies additional code data for each class code to a tap decreasing code 10. The additional code data is used to decrease taps of SD data. The tap decreasing circuit 10 decreases the SD data extracted by an area extracting circuit 9. A prediction calculating circuit 11 receives coefficient data corresponding to the class code from a ROM table 8 and obtains HD data with the decreased SD data corresponding to a linear prediction equation.

公开(公告)号：DE69228514T2

公开(公告)日：1999-08-12

申请号：DE69228514

申请日：1992-10-15

Applicant: SONY CORP

Inventor： KONDO TETSUJIRO ,  FUJIMORI YASUHIRO ,  UCHIDA MASASHI ,  HORISHI TAKASHI ,  HASHINO TSUKASA

IPC: G06F3/00 ,  G06T7/20 ,  H04N5/232

Abstract: In an apparatus for judging a hand movement of an image, pixels where absolute values of differences between image data of respective pixels of blocks of a current frame and image data of representative point pixels of blocks of an earlier frame read out from a representative point memory satisfy a predetermined condition are judged by a condition judgment circuit to form a frequency distribution table corresponding to the judged result by a frequency distribution table formation circuit to calculate, on the basis of the frequency distribution table, the number of coordinates having a frequency value greater than a predetermined value in the vicinity of a coordinate designated by a motion vector of an image to judge the motion vector to result from a hand movement when the calculated value is less than a predetermined value.

公开(公告)号：DE69324486D1

公开(公告)日：1999-05-20

申请号：DE69324486

申请日：1993-12-09

Applicant: SONY CORP

Inventor： KONDO TETSUJIRO SONY CORPORATI ,  UCHIDA MASASHI ,  KAWAGUCHI KUNIO SONY CORPORATI

IPC: H04N7/01 ,  H04N7/12 ,  H04N7/26

Abstract: A digitally inputted image signal (SD signal) is converted into a high resolution digital image signal (HD signal) by using estimated values. The specific pixels of an object to be estimated are classified according to the one-, two-, or three-dimension level distribution of plural reference pixels of SD signal located near the specific pixels. The estimated values of the specific pixels are generated by a linear combination of the values of the reference values and the estimation coefficients determined in advance by learning. During the learning, a known HD signal and the SD signal formed from it are used to determine the estimation coefficients so as to minimize the sum of squares of the errors between the true values and the values estimated by the linear combination of the values of the surrounding SD signal reference pixels and the estimated coefficients. Not necessarily limited to the estimated coefficients, it may be possible to use, as the estimated values, representatives determined for every class in correspondence with the class of the inputted SD signal. The reference value of a block and the value normalized by the dynamic range DR are an example as the representatives.