-
公开(公告)号：US11238274B2
公开(公告)日：2022-02-01
申请号：US16622586
申请日：2017-12-27
发明人： Wenmin Wang , Yihao Zhang , Ronggang Wang , Ge Li , Shengfu Dong , Zhenyu Wang , Ying Li , Wen Gao
摘要： An image feature extraction method for person re-identification includes performing person re-identification by means of aligned local descriptor extraction and graded global feature extraction; performing the aligned local descriptor extraction by processing an original image by affine transformation and performing a summation pooling operation on image block features of same regions to obtain an aligned local descriptor; reserving spatial information between inner blocks of the image for the aligned local descriptor; and performing the graded global feature extraction by grading a positioned pedestrian region block and solving a corresponding feature mean value to obtain a global feature. The method can resolve the problem of feature misalignment caused by posture changes of pedestrian, etc., and eliminate the effect of unrelated backgrounds on re-recognition, thus improving the precision and robustness of person re-identification.
-
公开(公告)号：US10789505B2
公开(公告)日：2020-09-29
申请号：US15631906
申请日：2017-06-23
申请人： Google LLC
发明人： Ofer Shacham , David Patterson , William R. Mark , Albert Meixner , Daniel Frederic Finchelstein , Jason Rupert Redgrave
摘要： A method is described that includes executing a convolutional neural network layer on an image processor having an array of execution lanes and a two-dimensional shift register. The executing of the convolutional neural network includes loading a plane of image data of a three-dimensional block of image data into the two-dimensional shift register. The executing of the convolutional neural network also includes performing a two-dimensional convolution of the plane of image data with an array of coefficient values by sequentially: concurrently multiplying within the execution lanes respective pixel and coefficient values to produce an array of partial products; concurrently summing within the execution lanes the partial products with respective accumulations of partial products being kept within the two dimensional register for different stencils within the image data; and, effecting alignment of values for the two-dimensional convolution within the execution lanes by shifting content within the two-dimensional shift register array.
-
公开(公告)号：US10776902B2
公开(公告)日：2020-09-15
申请号：US16427103
申请日：2019-05-30
发明人： Tobias Bergen , Michaela Benz , Andreas Ernst , Thomas Wittenberg , Christian Muenzenmayer , Frederik Zilly , Malte Avenhaus
IPC分类号： H04N5/232 , G06T3/40 , G06T5/00 , G06T5/50 , G06K9/56 , G06K9/46 , G06K9/62 , G06T7/33 , H04N5/225 , G06K9/20
摘要： Image processing device for producing in real-time a digital composite image from a sequence of digital images recorded by a camera device, in particular an endoscopic camera device, the image processing device including a selecting unit, a key point detection unit, a transforming unit and a joining unit,wherein the key point detection unit includes a maximum detection unit configured for executing following steps separately for the filter response for the reference image and for the filter response for the further image, wherein a variable threshold is used: i) creating blocks by dividing the respective filter response, ii) calculating the variable threshold for each of the blocks, iii) discarding those blocks of the blocks from further consideration, in which the respective filter response at a reference point of the respective block is less than the respective variable threshold.
-
公开(公告)号：US10546211B2
公开(公告)日：2020-01-28
申请号：US15201204
申请日：2016-07-01
申请人： Google LLC
发明人： Ofer Shacham , David Patterson , William R. Mark , Albert Meixner , Daniel Frederic Finchelstein , Jason Rupert Redgrave
摘要： A method is described that includes executing a convolutional neural network layer on an image processor having an array of execution lanes and a two-dimensional shift register. The two-dimensional shift register provides local respective register space for the execution lanes. The executing of the convolutional neural network includes loading a plane of image data of a three-dimensional block of image data into the two-dimensional shift register. The executing of the convolutional neural network also includes performing a two-dimensional convolution of the plane of image data with an array of coefficient values by sequentially: concurrently multiplying within the execution lanes respective pixel and coefficient values to produce an array of partial products; concurrently summing within the execution lanes the partial products with respective accumulations of partial products being kept within the two dimensional register for different stencils within the image data; and, effecting alignment of values for the two-dimensional convolution within the execution lanes by shifting content within the two-dimensional shift register array.
-
公开(公告)号：US20190332850A1
公开(公告)日：2019-10-31
申请号：US15992945
申请日：2018-05-30
申请人： Apple Inc.
发明人： Vinay Sharma , Yaozong Gao
摘要： Training a generative adversarial network (GAN) for use in facial recognition, comprising providing an input image of a particular face into a facial recognition system to obtain a faceprint; obtaining, based on the input faceprint and a noise value, a set of output images from a GAN generator; obtaining feedback from a GAN discriminator, wherein obtaining feedback comprises inputting each output image into the GAN discriminator and determining a set of likelihood values indicative of whether each output image comprises a facial image; determining, based on each output image, a modified noise value; inputting each output image into a second facial recognition network to determine a set of modified faceprints; defining, based on each modified noise value and modified faceprint, feedback for the GAN generator, wherein the feedback comprises a first value and a second value; and modifying control parameters of the GAN generator.
-
公开(公告)号：US10262185B2
公开(公告)日：2019-04-16
申请号：US15581247
申请日：2017-04-28
发明人： Po-Ching Chen , Cheng-Hsiang Ho
摘要： An image processing method includes obtaining a sensed image, wherein the sensed image comprises a pattern; dividing the sensed image into a plurality of blocks; calculating a direction field according to the pattern in each of the blocks; calculating a similarity degree between the direction field of a first block and the direction fields of adjacent blocks of the first block; and classifying the first block into a first part according to the similarity degree of the first block.
-
7.发明申请公开(公告)号：US20170061580A1
公开(公告)日：2017-03-02
申请号：US15088764
申请日：2016-04-01
发明人： Daniel Stan , Min Li , Vladimir Lachine
CPC分类号： G06T3/4007 , G06K9/4647 , G06K9/56 , G06T3/4015 , G06T3/403
摘要： Rescaling or reconstructing of a digital image may be accomplished by directional interpolation, so that interpolation is done in the direction perpendicular to the gradient—the direction in which the change in pixel values is the smallest. Each pixel is generated by interpolation in the output image as a weighted average of nearby pixels, in which the weighting is done in the direction of the gradient. The interpolation is accomplished with an adaptive filter that has an elliptical frequency response determined by the direction of the gradient. The filter uses filter coefficients that are a function of the direction. Rather than storing coefficients for each of several directions, three filter coefficients are stored—one set for non-directional filter, one for one direction such as 45 degrees, and another for another direction such as 135 degrees. A blending of the filter coefficients is used.
摘要翻译： 数字图像的重新缩放或重建可以通过方向插值来实现，使得在垂直于梯度的方向（像素值的变化最小的方向）上进行插值。 通过在输出图像中的内插生成每个像素作为附近像素的加权平均值，其中在渐变方向上进行加权。 使用具有由梯度方向确定的椭圆频率响应的自适应滤波器来完成内插。 滤波器使用作为方向函数的滤波器系数。 不是存储几个方向中的每一个的系数，而是存储三个滤波器系数，一个用于无方向滤波器，一个针对一个方向，例如45度，另一个用于另一个方向，例如135度。 使用滤波器系数的混合。
-
公开(公告)号：US20170053406A1
公开(公告)日：2017-02-23
申请号：US15137350
申请日：2016-04-25
申请人： EyeVerify Inc.
发明人： David Hirvonen
CPC分类号： H04N5/2256 , G06K9/00221 , G06K9/00906 , G06K9/56 , G06T7/32 , G06T7/70 , G06T7/90 , G06T2207/20024 , G06T2207/30196 , H04L63/0861 , H04N5/23293
摘要： In a system for determining liveness of an image presented for authentication, a reference signal is rendered on a display, and a reflection of the rendered signal from a target is analyzed to determine liveness thereof. The analysis includes spatially and/or temporally band pass filtering the reflected signal, and determining RGB values for each frame in the reflected signal and/or each pixel in one or more frames of the reflected signal. Frame level and/or pixel-by-pixel correlations between the determined RGB values and the rendered signal are computed, and a determination of whether an image presented is live or fake is made using either or both correlations.
摘要翻译： 在用于确定呈现用于认证的图像的活动性的系统中，在显示器上呈现参考信号，并且分析来自目标的渲染信号的反射以确定其活跃性。 分析包括对反射信号进行空间和/或时间带通滤波，并且在反射信号的一个或多个帧中确定反射信号和/或每个像素中的每个帧的RGB值。 计算所确定的RGB值和渲染信号之间的帧级和/或逐像素相关性，并且使用两者之一或两者相关来确定呈现的图像是活的还是假的。
-
公开(公告)号：US09330332B2
公开(公告)日：2016-05-03
申请号：US14046194
申请日：2013-10-04
CPC分类号： G06K9/4633 , G06K9/6247
摘要： An approach to computation of kernel descriptors is accelerated using precomputed tables. In one aspect, a fast algorithm for kernel descriptor computation that takes O(1) operations per pixel in each patch, based on pre-computed kernel values. This speeds up the kernel descriptor features under consideration, to levels that are comparable with D-SIFT and color SIFT, and two orders of magnitude faster than STIP and HoG3D. In some examples, kernel descriptors are applied to extract gradient, flow and texture based features for video analysis. In tests of the approach on a large database of internet videos used in the TRECVID MED 2011 evaluations, the flow based kernel descriptors are up to two orders of magnitude faster than STIP and HoG3D, and also produce significant performance improvements. Further, using features from multiple color planes produces small but consistent gains.
摘要翻译： 使用预先计算的表加速了内核描述符的计算方法。 在一个方面，一种用于内核描述符计算的快速算法，其基于预先计算的内核值在每个补丁中每像素执行O（1）个操作。 这将加速考虑的内核描述符功能，达到与D-SIFT和颜色SIFT相当的水平，比STIP和HoG3D快两个数量级。 在一些示例中，内核描述符被应用于提取用于视频分析的梯度，流和纹理的特征。 在对TRECVID MED 2011评估中使用的大量互联网视频数据库的方法进行测试时，基于流的内核描述符比STIP和HoG3D快两个数量级，并且还可以显着提高性能。 此外，使用来自多个颜色平面的特征产生小但恒定的增益。
-
公开(公告)号：US09189834B2
公开(公告)日：2015-11-17
申请号：US14080659
申请日：2013-11-14
发明人： Zhe Lin , Jianchao Yang , Hailin Jin , Xin Lu
CPC分类号： G06T5/002 , G06K9/4609 , G06K9/4642 , G06K9/56 , G06T5/50 , G06T2207/20012 , G06T2207/20021 , G06T2207/20182
摘要： In techniques for adaptive denoising with internal and external patches, example image patches taken from example images are grouped into partitions of similar patches, and a partition center patch is determined for each of the partitions. An image denoising technique is applied to image patches of a noisy image to generate modified image patches, and a closest partition center patch to each of the modified image patches is determined. The image patches of the noisy image are then classified as either a common patch or a complex patch of the noisy image, where an image patch is classified based on a distance between the corresponding modified image patch and the closest partition center patch. A denoising operator can be applied to an image patch based on the classification, such as applying respective denoising operators to denoise the image patches that are classified as the common patches of the noisy image.
摘要翻译： 在使用内部和外部补丁进行自适应去噪的技术中，从示例图像获取的示例图像修补程序分组到类似修补程序的分区中，并为每个分区确定分区中心修补程序。 将图像去噪技术应用于噪声图像的图像补丁以产生修改后的图像斑块，并确定每个修改后的图像斑块的最接近的分割中心斑块。 然后，噪声图像的图像块被分类为噪声图像的公共补丁或复杂补丁，其中基于相应修改的图像补丁和最接近的分割中心补丁之间的距离对图像补丁进行分类。 可以基于分类将去噪算子应用于图像补片，例如应用相应的去噪算子去除被分类为噪声图像的公共斑块的图像斑块。
-
-
-
-
-
-
-
-
-
搜索结果
337 条记录 (耗时 0.028 秒)
