-
1.发明申请公开(公告)号：US20190065817A1
公开(公告)日：2019-02-28
申请号：US15690037
申请日：2017-08-29
Applicant: KONICA MINOLTA LABORATORY U.S.A., INC.
Inventor： Maral Mesmakhosroshahi , Shubham Agarwal , Yongmian Zhang
Abstract: An artificial neural network system implemented on a computer for cell segmentation and classification of biological images. It includes a deep convolutional neural network as a feature extraction network, a first branch network connected to the feature extraction network to perform cell segmentation, and a second branch network connected to the feature extraction network to perform cell classification using the cell segmentation map generated by the first branch network. The feature extraction network is a modified VGG network where each convolutional layer uses multiple kernels of different sizes. The second branch network takes feature maps from two levels of the feature extraction network, and has multiple fully connected layers to independently process multiple cropped patches of the feature maps, the cropped patches being located at a centered and multiple shifted positions relative to the cell being classified; a voting method is used to determine the final cell classification.
-
Patent Agency Ranking
公开(公告)号：US10521697B2
公开(公告)日：2019-12-31
申请号：US15721610
申请日：2017-09-29
Applicant: KONICA MINOLTA LABORATORY U.S.A., INC.
Inventor： Shubham Agarwal , Maral Mesmakhosroshahi , Yongmian Zhang
Abstract: A local connectivity feature transform (LCFT) is applied to binary document images containing text characters, to generate transformed document images which are then input into a bi-directional Long Short Term Memory (LSTM) neural network to perform character/word recognition. The LCFT transformed image is a gray scale image where the pixel values encode local pixel connectivity information of corresponding pixels in the original binary image. The transform is one that provides a unique transform score for every possible shape represented as a 3×3 block. In one example, the transform is computed using a 3×3 weight matrix that combines bit coding with a zigzag pattern to assign weights to each element of the 3×3 block, and by summing up the weights for the non-zero elements of the 3×3 block shape.
-
3.发明申请公开(公告)号：US20190102653A1
公开(公告)日：2019-04-04
申请号：US15721610
申请日：2017-09-29
Applicant: KONICA MINOLTA LABORATORY U.S.A., INC.
Inventor： Shubham Agarwal , Maral Mesmakhosroshahi , Yongmian Zhang
Abstract: A local connectivity feature transform (LCFT) is applied to binary document images containing text characters, to generate transformed document images which are then input into a bi-directional Long Short Term Memory (LSTM) neural network to perform character/word recognition. The LCFT transformed image is a gray scale image where the pixel values encode local pixel connectivity information of corresponding pixels in the original binary image. The transform is one that provides a unique transform score for every possible shape represented as a 3×3 block. In one example, the transform is computed using a 3×3 weight matrix that combines bit coding with a zigzag pattern to assign weights to each element of the 3×3 block, and by summing up the weights for the non-zero elements of the 3×3 block shape.
-
公开(公告)号：US10762377B2
公开(公告)日：2020-09-01
申请号：US16236498
申请日：2018-12-29
Applicant: KONICA MINOLTA LABORATORY U.S.A., INC.
Inventor： Maral Mesmakhosroshahi
Abstract: A method processes scanned floating forms by comparing topological structures of an empty form and a corresponding filled form. The topological structure of each form includes vertical and horizontal ordered sequences of text phrases in the form, which describe directional relationships among text phrases but not their distances. A partial structure alignment method is used in the comparison, where insertions of text in the topological structure of the filled form relative to that of the empty form is not penalized, but deletions and substitutions are penalized. Based on this comparison, some text phrases in the filled form are matched to the those in the empty form, and unmatched text in the filled form is deemed user-filled data. The method further associates user-filled data with fields of the form based on a settings file of the empty form.
-
公开(公告)号：US20200210746A1
公开(公告)日：2020-07-02
申请号：US16236498
申请日：2018-12-29
Applicant: KONICA MINOLTA LABORATORY U.S.A., INC.
Inventor： Maral Mesmakhosroshahi
Abstract: A method processes scanned floating forms by comparing topological structures of an empty form and a corresponding filled form. The topological structure of each form includes vertical and horizontal ordered sequences of text phrases in the form, which describe directional relationships among text phrases but not their distances. A partial structure alignment method is used in the comparison, where insertions of text in the topological structure of the filled form relative to that of the empty form is not penalized, but deletions and substitutions are penalized. Based on this comparison, some text phrases in the filled form are matched to the those in the empty form, and unmatched text in the filled form is deemed user-filled data. The method further associates user-filled data with fields of the form based on a settings file of the empty form.
-
6.发明授权公开(公告)号：US10282589B2
公开(公告)日：2019-05-07
申请号：US15690037
申请日：2017-08-29
Applicant: KONICA MINOLTA LABORATORY U.S.A., INC.
Inventor： Maral Mesmakhosroshahi , Shubham Agarwal , Yongmian Zhang
Abstract: An artificial neural network system implemented on a computer for cell segmentation and classification of biological images. It includes a deep convolutional neural network as a feature extraction network, a first branch network connected to the feature extraction network to perform cell segmentation, and a second branch network connected to the feature extraction network to perform cell classification using the cell segmentation map generated by the first branch network. The feature extraction network is a modified VGG network where each convolutional layer uses multiple kernels of different sizes. The second branch network takes feature maps from two levels of the feature extraction network, and has multiple fully connected layers to independently process multiple cropped patches of the feature maps, the cropped patches being located at a centered and multiple shifted positions relative to the cell being classified; a voting method is used to determine the final cell classification.
-
-
-
-
-
Search Results
6
records
(took 0.013 seconds)
