公开(公告)号：US20050280739A1

公开(公告)日：2005-12-22

申请号：US10871729

申请日：2004-06-17

Applicant: Peng Lin ,  Yeong-Taeg Kim

Inventor： Peng Lin ,  Yeong-Taeg Kim

IPC: G06T5/00 ,  H04N5/14 ,  H04N5/21

CPC classification number: H04N5/144 ,  G06T5/002 ,  G06T5/20 ,  G06T5/50 ,  G06T2207/10016 ,  G06T2207/20182 ,  G06T2207/20192 ,  G06T2207/20216 ,  H04N5/21

Abstract: A video noise reduction system for a set of video frames that computes a first motion signal using a current frame and multiple consecutive previous frames, computes a second motion signal using the current frame and the processed preceding frame; computes the multi-frame temporal average of the current frame and multiple consecutive previous frames; computes the recursive average of the current frame and the processed preceding frame; generates a temporal filtered signal by soft switching between the multi-frame temporal average and the recursive average based on the first motion signal; applies a spatial filter to the current frame to generate a spatial filtered signal; and combines the temporal filtered signal and the spatial filtered signal based on the second motion signal to generate a final noise reduced video output signal.

Abstract translation: 一种用于使用当前帧和多个连续先前帧计算第一运动信号的一组视频帧的视频噪声降低系统，使用当前帧和处理的先前帧计算第二运动信号; 计算当前帧和多个连续的先前帧的多帧时间平均值; 计算当前帧和处理的前一帧的递归平均值; 通过基于第一运动信号的多帧时间平均和递归平均之间的软切换来产生时间滤波信号; 对当前帧应用空间滤波器以产生空间滤波信号; 并且基于所述第二运动信号组合所述时间滤波信号和所述空间滤波信号以生成最终降噪视频输出信号。

公开(公告)号：US08571100B2

公开(公告)日：2013-10-29

申请号：US11666061

申请日：2005-10-19

Applicant: Xiaoan Lu ,  Peng Lin ,  Jill MacDonald Boyce

Inventor： Xiaoan Lu ,  Peng Lin ,  Jill MacDonald Boyce

IPC: H04B1/66

CPC classification number: H04N19/19 ,  H04N19/109 ,  H04N19/11 ,  H04N19/14 ,  H04N19/147 ,  H04N19/176 ,  H04N19/593 ,  H04N19/61

Abstract: There are provided video encoders and corresponding methods for encoding video data for an image that is divisible into macroblocks. A video encoder includes an encoder for performing into mode selection when encoding a current macroblock by testing a first subset of intra modes to compute a rate distortion cost, and utilizing the rate distortion cost to determine whether to terminate the intra mode selection and which additional intra modes, if any, are to be examined with respect to the current macroblock.

Abstract translation: 提供了视频编码器和用于编码可分割为宏块的图像的视频数据的相应方法。 视频编码器包括编码器，用于通过测试帧内模式的第一子集来编码当前宏块进行模式选择，以计算速率失真成本，并利用速率失真成本来确定是否终止帧内模式选择以及哪些附加帧内 要对当前的宏块进行检查，如果有的话。

公开(公告)号：US08405750B2

公开(公告)日：2013-03-26

申请号：US12480651

申请日：2009-06-08

Applicant: Scott Smith ,  Peng Lin ,  Christopher Zeleznik ,  Laura Savidge

Inventor： Scott Smith ,  Peng Lin ,  Christopher Zeleznik ,  Laura Savidge

IPC: H04N5/335 ,  G03B7/00

CPC classification number: H04N9/045 ,  H04N5/235 ,  H04N5/2355 ,  H04N2209/046

Abstract: High dynamic range image sensors and image reconstruction methods for capturing high dynamic range images. An image sensor that captures high dynamic range images may include an array of pixels having two sets of pixels, each of which is used to capture an image of a scene. The two sets of pixels may be interleaved together. As an example, the first and second sets of pixels may be formed in odd-row pairs and even-row pairs of the array, respectively. The first set of pixels may use a longer exposure time than the second set of pixels. The exposures of the two sets of pixels may at least partially overlap in time. Image processing circuitry in the image sensors or an associated electronic device may de-interlace the two images and may combine the de-interlaced images to form a high dynamic range image.

Abstract translation: 高动态范围图像传感器和用于捕获高动态范围图像的图像重建方法。 捕获高动态范围图像的图像传感器可以包括具有两组像素的像素阵列，每组像素用于捕获场景的图像。 两组像素可以交错在一起。 作为示例，第一和第二组像素可以分别形成为阵列的奇数行和偶数行。 第一组像素可以使用比第二组像素更长的曝光时间。 两组像素的曝光可能在时间上至少部分重叠。 图像传感器或相关联的电子设备中的图像处理电路可以对两个图像进行去隔行扫描，并且可以组合去隔行图像以形成高动态范围图像。

公开(公告)号：US20120274822A1

公开(公告)日：2012-11-01

申请号：US13170478

申请日：2011-06-28

Applicant: Scott Smith ,  Dong Li ,  Peng Lin

Inventor： Scott Smith ,  Dong Li ,  Peng Lin

IPC: H04N5/335

CPC classification number: H04N5/23254 ,  H04N5/2353 ,  H04N5/2355 ,  H04N5/3535 ,  H04N5/35554

Abstract: An imager includes an array of pixels arranged in rows and a control circuit for sequentially capturing first and second image frames from the array of pixels. The control circuit is configured to sequentially capture first and second pairs of adjacent rows of pixels during first and second exposure times, respectively, when capturing the first image frame. The control circuit is also configured to sequentially capture first and second pairs of adjacent rows of pixels during second and first exposure times, respectively, when capturing the second image frame. The first exposure times during the first and second frames are of similar duration; and the second exposure times during the first and second frames are of similar duration. The control circuit is configured to detect motion of an object upon combining the first and second image frames and, then, correct for the motion of the object.

Abstract translation: 成像器包括排列成行的像素阵列和用于从像素阵列顺序捕获第一和第二图像帧的控制电路。 控制电路被配置为在捕获第一图像帧时分别在第一和第二曝光时间期间顺序捕获第一和第二对相邻像素行对。 控制电路还被配置为在捕获第二图像帧时分别在第二和第一曝光时间期间顺序捕获第一和第二对相邻像素行。 在第一和第二帧期间的第一曝光时间具有相似的持续时间; 并且第一和第二帧期间的第二曝光时间具有相似的持续时间。 控制电路被配置为在组合第一和第二图像帧时检测对象的运动，然后校正对象的运动。

公开(公告)号：US20120194686A1

公开(公告)日：2012-08-02

申请号：US13112964

申请日：2011-05-20

Applicant: Peng Lin ,  Lingtao Jiang ,  Scott Smith

Inventor： Peng Lin ,  Lingtao Jiang ,  Scott Smith

IPC: H04N5/228

CPC classification number: H04N5/2355

Abstract: Electronic devices may include image sensors and processing circuitry. Image sensors may be used to capture multiple exposure images. Processing circuitry may be used to combine multiple exposure images into high-dynamic-range images. A motion correction method is provided that detects motion between multiple exposure images without using a frame buffer. A noise model is used to separate noise from motion for more accurate motion detection. A dilation operator may be used to enlarge a motion mask generated by the motion detector. Motion-corrected images may be generated from the multiple exposure images using a soft switch based on the motion strength. Motion-corrected multiple exposure images may be combined to generate a motion-corrected HDR image. A smoothing filter may be applied to the motion region of the motion-corrected HDR image. A blooming correction may be used to eliminate color artifacts in the motion-corrected HDR image.

Abstract translation: 电子设备可以包括图像传感器和处理电路。 图像传感器可用于捕获多个曝光图像。 处理电路可以用于将多个曝光图像组合成高动态范围图像。 提供一种运动校正方法，其检测多个曝光图像之间的运动，而不使用帧缓冲器。 噪声模型用于将噪声与运动分离以进行更精确的运动检测。 扩张操作者可以用于放大由运动检测器产生的运动掩模。 可以使用基于运动强度的软开关从多个曝光图像生成运动校正图像。 运动校正的多个曝光图像可以被组合以产生运动校正的HDR图像。 可以将平滑滤波器应用于运动校正的HDR图像的运动区域。 可以使用开花校正来消除运动校正的HDR图像中的色彩伪影。

公开(公告)号：US20120170843A1

公开(公告)日：2012-07-05

申请号：US12985310

申请日：2011-01-05

Applicant: Peng Lin

Inventor： Peng Lin

IPC: G06K9/40 ,  G06K9/00

CPC classification number: G06T5/008 ,  G06T2207/10016 ,  G06T2207/20012 ,  G06T2207/20208

Abstract: Adaptive local tone mapping may be used to convert a high dynamic range image to a low dynamic range image. Tone mapping may be performed on an on a Bayer domain image. A high dynamic range image may be filtered to produce a luminance signal. An illumination component of the luminance signal may be compressed. A reflectance component of the luminance signal may be sharpened. After the luminance signal has been processed, it may be used in producing an output image in the Bayer domain that has a lower dynamic range than the input image. The output Bayer domain image may be demosaiced to produce an RGB image. Tone-mapping may be performed with a tone-mapping processor.

Abstract translation: 自适应局部色调映射可用于将高动态范围图像转换为低动态范围图像。 色调映射可以在拜耳域图像上执行。 可以对高动态范围图像进行滤波以产生亮度信号。 可以压缩亮度信号的照明分量。 亮度信号的反射分量可以被削尖。 在亮度信号被处理之后，它可以用于产生具有比输入图像低的动态范围的拜耳域中的输出图像。 输出拜耳域图像可以被去马赛克以产生RGB图像。 音调映射可以用色调映射处理器来执行。

公开(公告)号：US08184705B2

公开(公告)日：2012-05-22

申请号：US12145789

申请日：2008-06-25

Applicant: Peng Lin

Inventor： Peng Lin

IPC: H04N7/12 ,  H04N11/04

CPC classification number: H04N19/117 ,  H04N19/139 ,  H04N19/176 ,  H04N19/577 ,  H04N19/85

Abstract: A method and apparatus for filtering video data. First and second frames of the video data are stored. Motion adapted spatio-temporal filter (MASTF) pixel values of the second frame are calculated using pixel values of the first and second frames. The second frame is compared to the first frame to estimate motion vectors (MVs) for the second frame. Pixel MV error level indicators for the second frame are determined using the pixel values of the first and second frames and the MVs of pixels in the second frame. Motion compensated temporal filter (MCTF) pixel values of the second frame are calculated using the pixel values of the first and second frames and the MVs of pixels in the second frame. For each pixel in the second frame, a filtered pixel value is calculated using its MASTF and MCTF pixel values and its pixel MV error level indicator.

Abstract translation: 一种用于过滤视频数据的方法和装置。 存储视频数据的第一帧和第二帧。 使用第一和第二帧的像素值来计算第二帧的运动适应时空滤波器（MASTF）像素值。 将第二帧与第一帧进行比较，以估计第二帧的运动矢量（MV）。 使用第二帧的像素值和第二帧中的像素的MV来确定第二帧的像素MV误差水平指示符。 使用第二帧的像素值和第二帧中的像素的MV来计算第二帧的运动补偿时间滤波器（MCTF）像素值。 对于第二帧中的每个像素，使用其MASTF和MCTF像素值及其像素MV误差电平指示符来计算滤波像素值。

公开(公告)号：US08081224B2

公开(公告)日：2011-12-20

申请号：US12149723

申请日：2008-05-07

Applicant: Peng Lin

Inventor： Peng Lin

IPC: H04N5/228 ,  H04N5/235 ,  H04N5/262 ,  G06K9/40

CPC classification number: H04N5/23248 ,  H04N5/23267

Abstract: A method and apparatus for image stabilization while mitigating the amplification of image noise by using a motion adaptive system employing spatial and temporal filtering of pixel signals from multiple captured frames of a scene.

Abstract translation: 一种用于图像稳定的方法和装置，同时通过使用对来自场景的多个拍摄帧的像素信号进行空间和时间滤波的运动自适应系统来减轻图像噪声的放大。

公开(公告)号：US07808558B2

公开(公告)日：2010-10-05

申请号：US11799345

申请日：2007-04-30

Applicant: Peng Lin ,  Yeong-Taeg Kim

Inventor： Peng Lin ,  Yeong-Taeg Kim

IPC: H04N5/00 ,  H04N5/21 ,  H04N9/68 ,  G06K9/20 ,  G06K9/40

CPC classification number: H04N9/646

Abstract: A superior Color Transient Improvement technique is adaptive to the local image features, so that more natural color edge transition improvement can be accomplished. A gain control function is provided that depends on the local image feature so that different regions of the image can be treated differently. Further, a correction signal is controlled in such a way (by the local image feature) that neither undershoot nor overshoot occurs, eliminating the need for post-processing for undershoot/overshoot removal.

Abstract translation: 优异的色彩瞬态改进技术适用于局部图像特征，从而可以实现更自然的色彩边缘转换改进。 提供了取决于本地图像特征的增益控制功能，使得图像的不同区域可以被不同地对待。 此外，校正信号以这样的方式（通过局部图像特征）被控制，既不会发生下冲也不产生超调，因此不需要后处理用于下冲/过冲去除。

公开(公告)号：US20090323808A1

公开(公告)日：2009-12-31

申请号：US12145789

申请日：2008-06-25

Applicant: Peng Lin

Inventor： Peng Lin

IPC: H04N7/26

CPC classification number: H04N19/117 ,  H04N19/139 ,  H04N19/176 ,  H04N19/577 ,  H04N19/85

Abstract: A method and apparatus for filtering video data. First and second frames of the video data are stored. Motion adapted spatio-temporal filter (MASTF) pixel values of the second frame are calculated using pixel values of the first and second frames. The second frame is compared to the first frame to estimate motion vectors (MVs) for the second frame. Pixel MV error level indicators for the second frame are determined using the pixel values of the first and second frames and the MVs of pixels in the second frame. Motion compensated temporal filter (MCTF) pixel values of the second frame are calculated using the pixel values of the first and second frames and the MVs of pixels in the second frame. For each pixel in the second frame, a filtered pixel value is calculated using its MASTF and MCTF pixel values and its pixel MV error level indicator.

Abstract translation: 一种用于过滤视频数据的方法和装置。 存储视频数据的第一帧和第二帧。 使用第一和第二帧的像素值来计算第二帧的运动适应时空滤波器（MASTF）像素值。 将第二帧与第一帧进行比较，以估计第二帧的运动矢量（MV）。 使用第二帧的像素值和第二帧中的像素的MV来确定第二帧的像素MV误差水平指示符。 使用第二帧的像素值和第二帧中的像素的MV来计算第二帧的运动补偿时间滤波器（MCTF）像素值。 对于第二帧中的每个像素，使用其MASTF和MCTF像素值及其像素MV误差电平指示符来计算滤波像素值。