公开(公告)号：EP1605705A1

公开(公告)日：2005-12-14

申请号：EP04425415.9

申请日：2004-06-07

Applicant: STMicroelectronics S.r.l.

Inventor： Vella, Filippo ,  Bruna, Arcangelo Ranieri ,  Buemi, Antonio Vincenzo

IPC: H04N7/34 ,  H04N7/28 ,  G06T9/00

CPC classification number: H04N19/119 ,  H04N19/124 ,  H04N19/136 ,  H04N19/17 ,  H04N19/176 ,  H04N19/51 ,  H04N19/94

Abstract: A particularly efficient and simple method for Bayer color filter array (CFA) compression requires practically negligible computational complexity and external memory requirement while yielding a substantially lossless performance of the compression-decompression processes.
For each color channel, the process comprises the step of

gathering Bayer pattern pixel values by pairs, each pair being composed by two successive pixels belonging to the channel along the scanning direction of the pixels of the image, thus each pair of values representing a current input vector;
calculating a predictor vector of the input vector in terms of the differences between the values defining the input vector and a pair of prediction values generated according to a certain criterion, for representing a prediction error;
quantizing each so calculated predictor vector according to a heavier or lighter degree of quantization depending on whether the predictor vector is representative of an area of relatively uniform color of the image or of an area of relatively abrupt changes of colors of the image;
generating a multibit code representative of the quantized predictor vector of the input vector according to a certain compression ratio.

公开(公告)号：EP1605705B1

公开(公告)日：2009-01-28

申请号：EP04425415.9

申请日：2004-06-07

Applicant: STMicroelectronics S.r.l.

Inventor： Vella, Filippo ,  Bruna, Arcangelo Ranieri ,  Buemi, Antonio Vincenzo

IPC: H04N7/34 ,  H04N7/28 ,  G06T9/00

CPC classification number: H04N19/119 ,  H04N19/124 ,  H04N19/136 ,  H04N19/17 ,  H04N19/176 ,  H04N19/51 ,  H04N19/94

公开(公告)号：EP1632902A1

公开(公告)日：2006-03-08

申请号：EP04020969.4

申请日：2004-09-03

Applicant: STMicroelectronics S.r.l. ,  STMicroelectronics Limited

Inventor： Smith, Stewart G. ,  Bruna, Arcangelo Ranieri ,  Vella, Filippo

IPC: G06T9/00

CPC classification number: H04N19/192 ,  H04N19/115 ,  H04N19/124 ,  H04N19/146 ,  H04N19/15 ,  H04N19/182

Abstract: A method for image compression of a set of image data (I) includes operating a quantization operation (20; 150; 230) on the image data (I), said quantization operation (20; 150; 230) including controlling a compression factor by applying a scaled quantization level ( ), obtained by multiplying a first quantization level (Q) by a gain factor (G; Gs; Gi), the gain factor (G; Gs; Gi) being updated (100; 1000; 110; 300) as a function of a bit per pixel value (bpp) of a compressed image (O). The update operation (1000; 110; 300) includes an iterative procedure (110) comprising at least one iteration step (Eq. 6; Eq. 7) that provides for updating a current gain factor (G n+1 ) as a function of a previous gain (G n ) used for performing a previous compression step and as a function of a ratio (G n ) of the bit per pixel value (bpp) of the compressed image (O) at the previous compression step to a target bit per pixel value (T).
Preferred application is in Joint Photographic Experts Group (JPEG) and digital still cameras.

Abstract translation: 一组图像数据（I）的图像压缩方法包括对图像数据（I）进行量化操作（20; 150; 230），所述量化操作（20; 150; 230）包括通过 通过将第一量化级别（Q）乘以增益因子（G; Gs; Gi），被更新的增益因子（G; Gs; Gi）获得的缩放量化电平（）（100; 1000; 110; 300 ）作为压缩图像（O）的像素值（bpp）的位的函数。 更新操作（1000; 110; 300）包括迭代过程（110），该迭代过程包括至少一个迭代步骤（等式6;等式7），该迭代步骤用于更新当前增益因子（G n + 1） 用于执行先前压缩步骤的先前增益（G n）和作为先前压缩步骤的压缩图像（O）的比特每像素值（bpp）的比率（G n）的函数的目标比特 每像素值（T）。 联合图像专家组（JPEG）和数码相机的首选应用。

公开(公告)号：EP1632902B1

公开(公告)日：2007-12-05

申请号：EP04020969.4

申请日：2004-09-03

Applicant: STMicroelectronics S.r.l. ,  STMicroelectronics Limited

Inventor： Smith, Stewart G. ,  Bruna, Arcangelo Ranieri ,  Vella, Filippo

IPC: G06T9/00

CPC classification number: H04N19/192 ,  H04N19/115 ,  H04N19/124 ,  H04N19/146 ,  H04N19/15 ,  H04N19/182

公开(公告)号：EP1377040A1

公开(公告)日：2004-01-02

申请号：EP02425403.9

申请日：2002-06-19

Applicant: STMicroelectronics S.r.l.

Inventor： Vella, Filippo ,  Mancuso, Massimo

IPC: H04N5/232 ,  H04N7/26

CPC classification number: H04N5/23264 ,  H04N5/145 ,  H04N5/23248 ,  H04N5/23254 ,  H04N19/17 ,  H04N19/527 ,  H04N19/55

Abstract: A method of stabilizing an image sequence, said method comprising the following phases:

estimating (Mot_Est) a first global motion vector ( absGMV [n,ref]) comprising a first motion component ( absGMV y [n,ref] ) in a predetermined direction that has associated with it a first respective amplitude and a first respective direction, said first vector being representative of the motion with respect to a reference image ( Img ref ) of a first image ( Img n ) consisting of a pixel matrix,
associating said first component with either a wanted motion or an unwanted motion,
compensating said first component when it is associated with an unwanted motion,
   characterized in that the association phase comprises a phase of comparing the first amplitude of said component with a threshold compensation value ( T hor c om p ) assigned to the predetermined direction, said first component being associated with an unwanted or wanted motion whenever the first amplitude is, respectively, smaller than said threshold value ( T hor c om p ) or greater than/equal to it.

Abstract translation: 一种稳定图像序列的方法，所述方法包括以下阶段：估计（Mot_Est）包括与其相关联的预定方向上的第一运动分量（absGMVyÄn，refu）的第一全局运动矢量（absGMVÄn，refÜ） 第一相应振幅和第一相应方向，所述第一矢量表示相对于由像素矩阵组成的第一图像（Imgn）的参考图像（Imgn）的运动，所述第一图像（Imgn）将所述第一分量与想要的运动或 当所述第一分量与不想要的运动相关联时补偿所述第一分量，其特征在于，所述关联阶段包括将所述分量的第一幅度与分配给所述预定方向的阈值补偿值（Thor_comp）进行比较的相位，所述第一分量 每当第一幅度分别小于所述阈值时与无用或想要的运动相关联 e（Thor_comp）或大于/等于它。

公开(公告)号：EP1376471A1

公开(公告)日：2004-01-02

申请号：EP02425402.1

申请日：2002-06-19

Applicant: STMicroelectronics S.r.l.

Inventor： Vella, Filippo ,  Mancuso, Massimo

IPC: G06T7/20

CPC classification number: G06T7/215

Abstract: A method (Mot_Est) of estimating a global motion vector representative of the motion of a first digital image ( Img n ) with respect to a second digital image ( Img n-1 ), the first and the second image forming part of a sequence of images and being made up of, respectively, a first and a second pixel matrix. The method estimates the global motion vector on the basis of the estimate of at least one motion vector ( F_GMV, B_GMV ) of at least one region (F_Ground, B_Ground) of the first image ( Img n ) representative of the motion of the at least one region from the first image to the second image and comprising phases of:

subdividing the at least one region of the first image into a plurality of pixel blocks,
assigning (Block_Weight) to each block of the plurality a respective weighting coefficient calculated on the basis of a respective inhomogeneity measure,
estimating (GMV_Select) the at least one motion vector ( F_GMV, B_GMV ) of said at least one region (F_Ground, B_Ground) on the basis of the weighting coefficients assigned to each block of the at least one region.

Abstract translation: 一种估计代表第一数字图像（Imgn）相对于第二数字图像（Imgn-1）的运动的全局运动矢量的方法（Mot_Est），图像序列的第一和第二图像形成部分;以及 分别由第一和第二像素矩阵构成。 该方法基于代表至少一个运动的第一图像（Imgn）的至少一个区域（F_Ground，B_Ground）的至少一个运动矢量（F_GMV，B_GMV）的估计来估计全局运动矢量 区域，其包括以下步骤：将所述第一图像的所述至少一个区域细分为多个像素块，向所述多个的每个块分配（Block_Weight）各自的加权系数，所述各个加权系数基于 相应的不均匀性度量，基于分配给至少一个区域的每个块的加权系数来估计（GMV_Select）所述至少一个区域（F_Ground，B_Ground）的至少一个运动矢量（F_GMV，B_GMV）。