公开(公告)号：US07590298B2

公开(公告)日：2009-09-15

申请号：US11402602

申请日：2006-04-12

申请人： Zhigang Fan ,  Ricardo de Queiroz ,  Reiner Eschbach ,  Shen-Ge Wang

发明人： Zhigang Fan ,  Ricardo de Queiroz ,  Reiner Eschbach ,  Shen-Ge Wang

IPC分类号： G06K9/46

CPC分类号： H04N19/86 ,  H04N19/117 ,  H04N19/136 ,  H04N19/18 ,  H04N19/63 ,  H04N19/80

摘要： A process for improving the reconstruction of images compressed by the wavelet-based JPEG2000 specification. The basic reconstruction is followed by an edge-preserving filter (such as a sigma filter), another transform to the wavelet domain, an adjustment in the wavelet domain to yield the same quantized values as the compressed input image, and, finally, an inverse wavelet transform to yield the improved image. The process reduces the artifacts in sharp text and graphics documents, yielding images that would otherwise require larger files.

摘要翻译： 一种改进基于小波JPEG2000规范压缩图像重建的过程。 基本重建之后是边缘保留滤波器（如西格玛滤波器），另一个转换为小波域，小波域中的调整产生与压缩输入图像相同的量化值，最后是反向 小波变换产生改进的图像。 该过程减少了锐利文本和图形文档中的伪影，产生了否则将需要较大文件的图像。

公开(公告)号：US20070242891A1

公开(公告)日：2007-10-18

申请号：US11402602

申请日：2006-04-12

申请人： Zhigang Fan ,  Ricardo de Queiroz ,  Reiner Eschbach ,  Shen-Ge Wang

发明人： Zhigang Fan ,  Ricardo de Queiroz ,  Reiner Eschbach ,  Shen-Ge Wang

IPC分类号： G06K9/36 ,  G06K9/46

CPC分类号： H04N19/86 ,  H04N19/117 ,  H04N19/136 ,  H04N19/18 ,  H04N19/63 ,  H04N19/80

摘要： A process for improving the reconstruction of images compressed by the wavelet-based JPEG2000 specification. The basic reconstruction is followed by an edge-preserving filter (such as a sigma filter), another transform to the wavelet domain, an adjustment in the wavelet domain to yield the same quantized values as the compressed input image, and, finally, an inverse wavelet transform to yield the improved image. The process reduces the artifacts in sharp text and graphics documents, yielding images that would otherwise require larger files.

摘要翻译： 一种改进基于小波JPEG2000规范压缩图像重建的过程。 基本重建之后是边缘保留滤波器（如西格玛滤波器），另一个转换为小波域，小波域中的调整产生与压缩输入图像相同的量化值，最后是反向 小波变换产生改进的图像。 该过程减少了锐利文本和图形文档中的伪影，产生了否则将需要较大文件的图像。

公开(公告)号：US06608928B1

公开(公告)日：2003-08-19

申请号：US09432600

申请日：1999-11-03

申请人： Ricardo de Queiroz

发明人： Ricardo de Queiroz

IPC分类号： G06K934

CPC分类号： G06K9/38 ,  G06T9/00 ,  G06T9/001 ,  H04N1/40062 ,  H04N1/41

摘要： A mixed raster content system and method improves the performance of the compression process by reducing the amount of data. The system and method uses a selector plane as a reference to aid in reducing the amount of data necessary to encode each of associated planes. A smoothing technique is used to pre-process the associated planes using the information contained in the selector plane, thereby reducing the amount of data that will be subjected to further processing. The system and method determines the useless data, smooths the boundary of the useless data, and replaces the useless data by values that improve compression.

摘要翻译： 混合光栅内容系统和方法通过减少数据量来提高压缩过程的性能。 该系统和方法使用选择器平面作为参考来帮助减少编码每个相关联平面所需的数据量。 使用平滑技术来使用包含在选择器平面中的信息对相关平面进行预处理，从而减少将被进一步处理的数据量。 系统和方法确定无用数据，平滑无用数据的边界，并通过改进压缩的值替换无用数据。

公开(公告)号：US20050041051A1

公开(公告)日：2005-02-24

申请号：US10644472

申请日：2003-08-20

申请人： Ricardo de Queiroz ,  Robert Loce ,  Karen Braun ,  Raja Bala

发明人： Ricardo de Queiroz ,  Robert Loce ,  Karen Braun ,  Raja Bala

IPC分类号： H04N1/32 ,  B41J2/205

CPC分类号： B41J2/205 ,  H04N1/32144 ,  H04N1/32251 ,  H04N1/32267 ,  H04N1/32309

摘要： What is disclosed is a method for digital watermarking in a calibrated printing path and comprises: first receiving a pixel possessing color values from an input image; receiving a plurality of information bits to be encoded at a corresponding pixel in an output image. Then, one of at least two different GCR functions are selected where the selection is based on the state of the received information bits. The number of GCR functions to be selected from is dependent on the number of possible states of the information bits intended to be encoded at each image pixel and preferably equals the number of states of the information bits intended to be encoded at each image pixel such that the GCR spatially varies across the output image. Further, at least two GCR functions are optimized to carry information and information bits intended to be encoded within the output image are represented with a tag. CMYK values are then generated using the selected GCR function and the color values. These CMYK values are assigned to a corresponding pixel in the output image. The information bits to be encoded at a given pixel indicate the type of object to which that pixel belongs, such as: graphics, picture, text, line art, etc. The output image, when printed, exhibits the property that substantially similar colors occurring at different spatial locations in the input image are produced with substantially different CMYK combinations in the print. Preferably, a parameterized function is used for the GCR function and the encoded state sets the parameter of the function. Information bits should be redundantly encoded throughout the output image. Regions that cannot be encoded by GCR information have to be compensated for. A reference mark is applied on the output image to indicate the starting point and order of the information sequence.

摘要翻译： 所公开的是在校准打印路径中进行数字水印的方法，包括：首先从输入图像接收具有颜色值的像素; 在输出图像中的相应像素处接收要编码的多个信息比特。 然后，选择至少两个不同的GCR功能中的一个，其中选择基于所接收的信息比特的状态。 要选择的GCR功能的数量取决于要在每个图像像素处被编码的信息比特的可能状态的数量，并且优选地等于要在每个图像像素处编码的信息比特的状态数，使得 GCR在输出图像之间空间变化。 此外，至少两个GCR功能被优化以携带信息，并且用标签表示旨在在输出图像内编码的信息位。 然后使用所选的GCR功能和颜色值生成CMYK值。 这些CMYK值被分配给输出图像中的相应像素。 在给定像素处要编码的信息位指示像素所属于的对象的类型，例如：图形，图片，文本，线条等。输出图像在打印时表现出基本相似的颜色出现的属性 在输入图像中的不同空间位置处，在打印中以基本不同的CMYK组合产生。 优选地，GCR函数使用参数化函数，编码状态设置函数的参数。 信息位应在整个输出图像中进行冗余编码。 不能由GCR信息编码的区域必须得到补偿。 在输出图像上应用参考标记来指示信息序列的起点和顺序。

公开(公告)号：US20050100223A1

公开(公告)日：2005-05-12

申请号：US10345004

申请日：2003-01-15

申请人： Ricardo de Queiroz

发明人： Ricardo de Queiroz

IPC分类号： G06K9/20 ,  G06K9/62

CPC分类号： G06K9/00456

摘要： This invention relates to a method and apparatus for image classification. More particularly, the present invention provides a technique to classify an image as a picture, a graphic or a mixed mode image. The classification is based on an approximation of a segmentation. The approximation is HVQ-LUT-based and outputs classification maps indicating whether pixels are background, text or pictures. Said classification maps are filtered to eliminate odd isolated samples and the resulting count of picture, text and background pixel is analyzed before concluding whether the image has pictorial, graphical or mixed contents.

摘要翻译： 本发明涉及图像分类的方法和装置。 更具体地，本发明提供了一种将图像分类为图片，图形或混合模式图像的技术。 分类基于分割的近似。 近似值是基于HVQ-LUT，并输出指示像素是背景，文本还是图片的分类图。 所述分类图被滤波以消除奇数分离的样本，并且在结束图像是否具有图形，图形或混合内容之前分析图像，文本和背景像素的结果计数。

公开(公告)号：US20050069197A1

公开(公告)日：2005-03-31

申请号：US10674048

申请日：2003-09-29

申请人： Ricardo de Queiroz ,  Karen Braun

发明人： Ricardo de Queiroz ,  Karen Braun

IPC分类号： G06T1/00 ,  G06K9/00 ,  G06K9/36 ,  H04N1/40 ,  H04N1/46 ,  H04N1/60

CPC分类号： H04N1/40012

摘要： A method is disclosed for converting color images to textured monochrome images such that image regions with similar luminance but different chrominance look different when converted to black-and-white to preserve color information therein. This texture-encoded color information can be extracted at a later time to reconstitute the original color image. The present method involves first converting the color image to a luminance-chrominance colorspace. The chrominance data is decomposed into 4 channels of chrominance. A wavelet transformation of the luminance channel is preformed. Scaled chrominance channels are then mapped to a number of wavelet sub-bands. Once mapped, the wavelet transform is inverted to generate textures proportional to the original colors of the color image having an amplitude proportional to the chroma of the original color. The black-and-white image is embedded with this texture information into the image's grayscale component. In order to retrieve the color image from the textured grayscale image, a wavelet transform regenerates the scaled chrominance channels mapped therein. Those wavelet sub-bands comprising embedded chroma are extracted and scaled to the size of the original image. The remaining sub-bands are inverted to produce the Y-image thereof. Separate images now exist in Y, Cr, and Cb. The recombination of the YCbCr images produce the colored image that was previously embedded into the grayscale components of the textured black-and-white image.

摘要翻译： 公开了一种将彩色图像转换为纹理单色图像的方法，使得具有相似亮度但是不同色度的图像区域在转换为黑白色时看起来不同，以保留其中的颜色信息。 可以在稍后时间提取该纹理编码的颜色信息以重构原始彩色图像。 本方法包括首先将彩色图像转换为亮度 - 色度色彩空间。 色度数据分解成4通道色度。 执行亮度通道的小波变换。 缩放的色度信道然后被映射到多个小波子带。 一旦被映射，小波变换被反转以产生与具有与原始颜色的色度成比例的幅度的彩色图像的原始颜色成比例的纹理。 黑白图像将该纹理信息嵌入到图像的灰度分量中。 为了从纹理灰度图像中检索彩色图像，小波变换再生映射在其中的缩放的色度通道。 提取包括嵌入色度的那些小波子带，并缩放到原始图像的大小。 剩余的子带被反转以产生其Y图像。 分离图像现在存在于Y，Cr和Cb中。 YCbCr图像的复合产生先前嵌入纹理黑白图像的灰度分量中的彩色图像。