专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

51. 发明申请

WO2020064253A1 METHODS FOR GENERATING A DEEP NEURAL NET AND FOR LOCALISING AN OBJECT IN AN INPUT IMAGE, DEEP NEURAL NET, COMPUTER PROGRAM PRODUCT, AND COMPUTER-READABLE STORAGE MEDIUM 审中-公开
公开(公告)号：WO2020064253A1
公开(公告)日：2020-04-02
申请号：PCT/EP2019/072960
申请日：2019-08-28
申请人： SIEMENS AKTIENGESELLSCHAFT
发明人： GHOSH, Sanjukta , AMON, Peter , HUTTER, Andreas
IPC分类号： G06K9/00 , G06K9/46 , G06K9/62
摘要： The invention relates to methods (1) for generating a deep neural net (10) and for localising an object (30) in an input image (2), the deep neural net (10), a corresponding computer program product, and a corresponding computer-readable storage medium. The invention proposes to train a discriminative counting model to classify images (2) according to a number of objects (30) of a predetermined type depicted in each of the images (2), and to train a segmentation model to segment images (2) by classifying each pixel according to what image part (30, 31, 35-38, 42, 60) the pixel belongs to. Parts (11) and/or features of both models are combined to form the deep neural net (10), wherein the deep neural net (10) is adapted to generate in a single forward pass a map (14, 16, 52, 58, 63) indicating locations of any objects (30) for each input image (2).

52. 发明申请

WO2018224442A1 METHOD AND APPARATUS FOR ANALYSING AN IMAGE 审中-公开
公开(公告)号：WO2018224442A1
公开(公告)日：2018-12-13
申请号：PCT/EP2018/064631
申请日：2018-06-04
申请人： SIEMENS AKTIENGESELLSCHAFT
发明人： GHOSH, Sanjukta , AMON, Peter , HUTTER, Andreas
IPC分类号： G06K9/00 , G06K9/62
摘要： The invention relates to a method (12, 17, 23) and an apparatus (15) for analysing an image (13, 18, 24, 48) using a deep neural net (25) pre-trained for multiple classes. The image (13, 18, 24, 48) is processed by means of a forward pass (20, 30) through an adapted neural net to generate a processing result. The adapted neural net is adapted from the pre-trained neural net (25) to focus on exactly one selected class. The processing result is then analysed focused on features corresponding to the selected class using an image processing algorithm. A modified image (55) is generated by removing a manifestation (49) of these features from the image (13, 18, 24, 48).

53. 发明申请

WO2015144381A1 VERFAHREN ZUR ÜBERTRAGUNG DIGITALER BILDER AUS EINER BILDERFOLGE 审中-公开
标题翻译：方法数字图像从图像传输成功
公开(公告)号：WO2015144381A1
公开(公告)日：2015-10-01
申请号：PCT/EP2015/054025
申请日：2015-02-26
申请人： SIEMENS AKTIENGESELLSCHAFT
发明人： AMON, Peter
IPC分类号： H04N21/845 , H04N21/2343 , H04N21/262 , H04N21/44 , H04N21/6587
CPC分类号： H04N21/238 , H04N21/23439 , H04N21/26258 , H04N21/44004 , H04N21/6587 , H04N21/8456
摘要： Die Erfindung betrifft ein Verfahren zur Übertragung digitaler Bilder (B, B') aus einer Bilderfolge, für welche eine Bildreihenfolge (BR) der Bilder (B, B') vorgegeben ist, wobei die Bilderfolge in der Form von Segmenten (SG1, SG2, SG3) auf einem Server (SE) hinterlegt ist und ein jeweiliges Segment (SG1, SG2, SG3) ein oder mehrere Bilder (B, B') umfasst, die gemäß der Bildreihenfolge (BR) benachbart sind, wobei ein Client (CL) Bilder (B, B') basierend auf den Segmenten (SG1, SG2, SG3) bei dem Server (SE) anfordert, der daraufhin die Bilder (B, B') aus den angeforderten Segmenten (SG1, SG2, SG3) zumindest teilweise zu dem Client (CL) übermittelt, in dem die Bilder (B, B') in einem Puffer (MB) gespeichert und zumindest zum Teil wiedergegeben werden. Der Client (CL) prüft zum Zeitpunkt der Wiedergabe eines aktuellen Bildes (B, B') das Anfordern von weiteren Bildern (B, B') vom Server (SE) basierend auf einer vorgegebenen Prioritätsregel prüft, wobei der Client (CL) im Rahmen der Prüfung des Anforderns eines jeweiligen weiteren Bildes (B, B') ein Segment (SG1, SG2, SG3) anfordert, welche das jeweilige weitere Bild (B, B') umfasst, falls das jeweilige weitere Bild (B, B') nicht im Puffer (MB) vorhanden ist. Die vorgegebene Prioritätsregel ist derart ausgestaltet, dass durch den Client (CL) zunächst das Anfordern von weiteren Bildern (B, B') geprüft wird, welche sowohl in Richtung der Bildreihenfolge (BR) als auch entgegen der Richtung der Bildreihenfolge (BR) auf das aktuelle Bild (B, B') folgen, und/oder zunächst das Anfordern zumindest eines weiteren Bildes (B, B') geprüft wird, welches dem aktuellen Bild (B, B') in einer höheren Bildqualität als die des aktuellen Bildes (B, B') entspricht.
摘要翻译：本发明涉及一种用于“从图像序列，用于将图像序列（BR）的图像的（B，B是预定的）传输的数字图像（B，B）”，其中，在段的形式的图像序列（SG1，SG2， SG3）（服务器SE）沉积和相应的段（SG1，SG2，SG3）一个或多个图像（B，B“），其BR）是相邻的（根据图像序列，其中一个客户机（CL）图像上 '基于所述段（SG1，SG2，SG3）发送到服务器（SE）请求，这然后将图像（B，B所请求的段）（SG1，SG2，SG3）至少部分地与（B，B）' 客户机（CL）发射，其中，所述图像被存储（B，B“）在缓冲器（MB）和至少部分地再生。客户机（CL）从所述服务器（SE）基于预定的优先级规则的检查，检查当前图像的再现（B，B“）请求其他图像（B，B”）的时间，其中，所述框架内的客户机（CL）请求相应的另外的图像（B，B“）的所述检查请求段（SG1，SG2，SG3），其相应的另外的图像（B，B”包括）如果相应的另外的图像（B，B“）是不在缓冲器（MB）的存在。预定的优先级规则被设计为使得由所述客户机（CL）第一请求附加图像（B，B“）被选中，这无论是在图像序列（BR）和相对的图像序列（BR）的方向的方向上的当前图像（B，B“）跟随，和/或第一请求至少一个另外的图像（B，B”），则检查当前图像（B，B“）（在更高的图像质量比当前图像B的对应于B“）。

54. 发明公开

EP2946550A1 A METHOD FOR CODING A SEQUENCE OF DIGITAL IMAGES 审中-公开
标题翻译：方法编码数字图像序列
公开(公告)号：EP2946550A1
公开(公告)日：2015-11-25
申请号：EP13745031.8
申请日：2013-07-26
申请人： Siemens Aktiengesellschaft
发明人： WIGE, Eugen , AMON, Peter , HUTTER, Andreas , KAUP, André
IPC分类号： H04N7/24
CPC分类号： H04N19/103 , G06K9/6215 , H04N19/105 , H04N19/11 , H04N19/124 , H04N19/14 , H04N19/167 , H04N19/176 , H04N19/44 , H04N19/46 , H04N19/593 , H04N19/61 , H04N19/65 , H04N19/70
摘要： A method for coding a sequence of digital images The invention refers to a method for coding a sequence of digital images (I), wherein the method uses a number of prediction modes for predicting values of pixels (P1) in the images (I) based on reconstructed values of pixels in image areas processed previously, where a prediction error (PE) between predicted values and the original values of pixels (P1) is processed for generating the coded sequence of digital images (CI). The invention is characterized in that a preset prediction mode is an intra-prediction mode based on pixels of a single image (I) and comprises steps i) and ii) as follows. In step i), for a region (R) of pixels with reconstructed values in the single image (I) and for a template (TE) of an image area, a first patch (PA1) of pixels in the region (R) which surround a first pixel (P1) to be predicted based on the template (TE) is compared with several second patches (PA2), each second patch (PA2) being assigned to a second pixel (P2) from a plurality of second pixels (P2) in the region (R) and consisting of pixels in the region (R) which surround the second pixel (P2) based on the template (TE), thereby determining a similarity measure (SM) for each second pixel (P2) describing the similarity between reconstructed values of the pixels of the second patch (PA2) assigned to the respective second pixel (P2) and the reconstructed values of the pixels of the first patch (PA1). In step ii), a predicted value of the first pixel (P1) is deter- mined based on the values of one or more second pixels (P2) which have the highest similarity described by the similarity measure (SM) among all second pixels (P2) of the plurality of second pixels (P2) in the region (R).

55. 发明授权

EP2489187B1 VERFAHREN ZUR CODIERUNG VON SYMBOLEN AUS EINER FOLGE DIGITALISIERTER BILDER 有权转让
标题翻译：方法用于序列数字化图像编码符号
公开(公告)号：EP2489187B1
公开(公告)日：2013-09-18
申请号：EP10765622.5
申请日：2010-10-13
申请人： Siemens Aktiengesellschaft
发明人： AMON, Peter
IPC分类号： H04N7/26 , H04N7/50
CPC分类号： H04N19/436 , H04N19/13 , H04N19/174

56. 发明公开

EP2617198A1 KOMPRESSION UND DEKOMPRESSION VON REFERENZBILDERN IN EINEM VIDEOKODER 审中-公开
标题翻译： EINEM VIDEOKODER中的压缩和压缩参考电压
公开(公告)号：EP2617198A1
公开(公告)日：2013-07-24
申请号：EP12705230.6
申请日：2012-01-12
申请人： Siemens Aktiengesellschaft
发明人： AGTHE, Bernhard , KUTKA, Robert , AMON, Peter , BÄSE, Gero , HUTTER, Andreas , OERTEL, Norbert
IPC分类号： H04N7/26
CPC分类号： H04N19/423 , H04N19/117 , H04N19/124 , H04N19/147 , H04N19/159 , H04N19/428 , H04N19/46
摘要： The invention relates to methods and devices for transforming image data, which are transformed by a compression filter before being compressed and stored in a reference image memory. In an extension, an inverse transformation to that of the compression filter is performed by a decompression filter when image data from the reference memory are read out and decompressed. The invention can be used for image compression methods and image decompression methods that use reference image memories.
摘要翻译：用于变换图像数据的方法和设备技术领域本发明涉及用于变换图像数据的方法和设备，该图像数据在被压缩并被存储在参考图像存储器中之前被压缩滤波器变换。在扩展中，当来自参考存储器的图像数据被读出并解压缩时，通过解压滤波器执行对压缩滤波器的逆变换。本发明可以用于使用参考图像存储器的图像压缩方法和图像解压缩方法。

57. 发明公开

EP2614647A1 KOMPRESSION UND DEKOMPRESSION VON REFERENZBILDERN IN EINEM VIDEOKODER 有权
标题翻译：在EINEM VIDEOKODER中的KOMPRESSION UND DEKOMPRESSION VON REFERENZBILDERN
公开(公告)号：EP2614647A1
公开(公告)日：2013-07-17
申请号：EP12700392.9
申请日：2012-01-12
申请人： Siemens Aktiengesellschaft
发明人： AMON, Peter , OERTEL, Norbert , BÄSE, Gero , HUTTER, Andreas , AGTHE, Bernhard , KUTKA, Robert
IPC分类号： H04N7/26
CPC分类号： H04N19/423 , H04N19/117 , H04N19/124 , H04N19/147 , H04N19/159 , H04N19/428 , H04N19/46
摘要： The invention relates to methods and devices for providing a reconstructed image within the framework of a predictive coding. According to the invention, the reconstructed image is compressed according to a coding mode of an image coding device and stored in a reference image memory.
摘要翻译：方法和装置将压缩滤波器变换的图像数据转换成压缩并存储在参考图像存储器中。在扩展中，当读出和解压缩来自参考存储器的图像数据时，通过解压缩滤波器执行与压缩滤波器的逆变换。方法和装置可用于使用参考图像存储器的图像压缩方法和图像解压缩方法。

58. 发明公开

EP2100455A2 VERFAHREN ZUR VIDEOCODIERUNG EINER FOLGE DIGITALISIERTER BILDER 审中-公开
标题翻译：方法视频编码的序列数字化图像
公开(公告)号：EP2100455A2
公开(公告)日：2009-09-16
申请号：EP07821324.6
申请日：2007-10-15
申请人： Siemens Aktiengesellschaft
发明人： AMON, Peter , PANDEL, Jürgen
IPC分类号： H04N7/26
CPC分类号： H04N19/67 , H04N19/114 , H04N19/177 , H04N19/31 , H04N19/577 , H04N19/66
摘要： The invention relates to a method for video-coding a sequence of digitized pictures. In said method: groups of pictures (GOP) are formed, each of which comprises a plurality of chronologically successive pictures (N0, P1, N2, I3, N4, P5, N6) in an original chronological order; each group of pictures (GOP) is coded by forming a prediction structure according to which one or more pictures of the group of pictures (GOP) is/are determined as intra-frames (I3), each of which is intra-coded, while the other pictures of the group of pictures (GOP) are determined as inter-frames (N0, P1, N2, N4, P5, N6), each of which is predicted from at least one reference frame of the group of pictures (GOP) and is inter-coded in relation to the at least one reference frame. The prediction structure is designed such that each intra-frame (I3) is a reference frame from which at least one picture (P1) of the group of pictures (GOP) that precedes the intra-frame (I3) as well as at least one picture (P5) of the group of pictures (GOP) that succeeds the intra-frame (I3) are predicted, the inter-frames (N0, P1, N2, N4, P5, N6) comprising several non-referenced pictures (N0, N6) from which no pictures of the sequence are predicted. A transmission sequence having a chronological transmission order is formed from the coded pictures (N0, P1, N2, I3, N4, P5, N6) of the group of pictures (GOP), at least some of the coded non-referenced pictures (N0, N1) being the first pictures of the transmission order.

59. 发明公开

EP1905038A1 VERFAHREN ZUR SPEICHERUNG EINZELNER DATENELEMENTE EINES SKALIERBAREN DATENSTROMS IN EINE DATEI SOWIE ZUGEHÖRIGE VORRICHTUNG 审中-公开
标题翻译：方法：针对可伸缩数据流的存储数据的单一稀土元素在文件中，相关设备
公开(公告)号：EP1905038A1
公开(公告)日：2008-04-02
申请号：EP06763760.3
申请日：2006-06-16
申请人： SIEMENS AKTIENGESELLSCHAFT
发明人： AMON, Peter , HUTTER, Andreas , RATHGEN, Thomas
IPC分类号： G11B27/10 , G11B27/32 , G11B27/30 , H04N5/76
CPC分类号： H04N9/8205 , G11B27/105 , G11B27/3027 , G11B27/322 , H04N5/765 , H04N5/77 , H04N9/8042 , H04N9/8227 , H04N21/2312 , H04N21/8451
摘要： The invention relates to a method for storing individual data elements of a scalable data flow in at least one file. Quality levels of the scalable data flow are described by at least one scaling feature in respective scaling levels and at least one data element is associated with every scaling level of the scaling features. A processing index is allocated to the data element in such a manner that only one or more data elements having a low value of the respective processing index have to be taken into consideration for processing. At least one descriptive list is produced in such a manner that it comprises descriptive elements associated with the data element, namely the scaling level of the respective scaling features, the time index and/or the processing index. At least one of the descriptive lists and the data elements associated therewith can be stored in one of the files in an organized manner. The invention also relates to a device for implementing and carrying out the method.

60. 发明公开

EP1829378A1 BILDENCODIERVERFAHREN, SOWIE DAZUGEHÖRIGES BILDDECODIERVERFAHREN, ENCODIERVORRICHTUNG UND DECODIERVORRICHTUNG 有权
标题翻译： BILDENCODIERVERFAHREN和相关联的图像进行解码的方法，和解码编码装置DEVICE
公开(公告)号：EP1829378A1
公开(公告)日：2007-09-05
申请号：EP05819040.6
申请日：2005-12-09
申请人： SIEMENS AKTIENGESELLSCHAFT
发明人： AMON, Peter , BÄSE, Gero , HUTTER, Andreas , PANDEL, Jürgen , TIMMERMANN, Benoit
IPC分类号： H04N7/26
CPC分类号： H04N19/615 , H04N19/107 , H04N19/114 , H04N19/13 , H04N19/61 , H04N19/63
摘要： Disclosed is an image encoding method in which a series of decoded images are generated from a series of original images with the aid of a first encoding process. Furthermore, a series of second images are generated from an image group of the series of original images with the aid of an INTER encoding mode of the second encoding process, at least one decoded image being used as a reference image. Said reference image represents an original image that is different from the original image represented by the second image of the image area that is to be encoded. The invention further relates to an image decoding method by means of which the second images generated by the image encoding method can be decoded. The invention finally relates to an encoding device and a decoding device by means of which the image encoding method or the image decoding method can be carried out.

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式