CN106650696B - method for identifying handwritten electrical element symbol based on singular value decomposition - Google Patents
method for identifying handwritten electrical element symbol based on singular value decomposition Download PDFInfo
- Publication number
- CN106650696B CN106650696B CN201611257061.1A CN201611257061A CN106650696B CN 106650696 B CN106650696 B CN 106650696B CN 201611257061 A CN201611257061 A CN 201611257061A CN 106650696 B CN106650696 B CN 106650696B
- Authority
- CN
- China
- Prior art keywords
- singular value
- value decomposition
- matrix
- training
- handwritten
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/32—Digital ink
- G06V30/333—Preprocessing; Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/32—Digital ink
- G06V30/36—Matching; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
A handwritten electric element symbol recognition method based on singular value decomposition comprises training and testing, wherein the training comprises the steps of carrying out binarization and denoising processing on an electric power engineering drawing, and cutting and standardizing elements and handwritten electric element symbols; training a singular value decomposition model by using a singular value decomposition idea; the testing includes cutting and standardizing the hand-written electrical element symbols; and identifying the elements through a trained singular value decomposition model. The characteristic extraction in the invention is to link the handwriting electric element with the standard electric element according to the related thought of singular value decomposition, and the characteristic extraction is favorable for the characteristic recognition of the handwriting electric element; and classifying the samples by adopting a nearest neighbor classifier, and judging the accuracy of the identification method.
Description
Technical Field
the invention relates to a handwritten electrical element symbol recognition method based on singular value decomposition, and belongs to the technical field of machine learning and pattern recognition.
background
With the continuous development of computer technology and the continuous enhancement of the computing processing capacity of computers, computers have been applied to many fields. In the aspect of electrical engineering, with the development of technology and the progress of times, drawing an electrical engineering drawing by using a computer is inevitable, however, a large number of hand-drawing drawings still exist in the existing electrical engineering drawing, if elements in the hand-drawing drawings are manually input into the computer, the input cost is greatly increased, and the time consumption is huge. Meanwhile, once the personnel entry is wrong, the influence on a project cannot be estimated due to the huge and fine workload. Therefore, the identification of the electrical elements in the electric engineering drawing is a necessary topic, which is beneficial to improving the efficiency of inputting the hand-drawn drawing into a computer, plays a promoting role in the automation of the electric engineering drawing, and lays a foundation for data statistics, data mining analysis and the like of relevant units such as a power grid, a research institute and the like.
Singular value decomposition is an important matrix decomposition process in the field of linear algebra, is popularization of unitary diagonalization of a normal matrix in matrix analysis, and has important application in the fields of digital signal processing, data statistics and the like. The singular value decomposition is similar in some respects to the diagonalization of symmetric or Hermite matrices based on eigenvectors, however, the two matrix decompositions are significantly different despite their correlation. The basis of the symmetric array eigenvector decomposition is spectral analysis, and the singular value decomposition is the popularization of the spectral analysis theory on any matrix. In the field of machine learning and pattern recognition, singular value decomposition is often applied to training of a sample set, and dimension reduction is performed on sample features, so that more important features are extracted, and classification recognition is promoted.
disclosure of Invention
Aiming at the defects in the prior art, the invention provides a handwritten electrical element symbol recognition method based on singular value decomposition. According to the method, the handwritten electric element symbols are recognized by machine learning and mode recognition related knowledge and the idea of singular value decomposition, so that the handwritten electric power engineering drawing is effectively utilized, the automation efficiency of the electric power engineering drawing is improved, and related work is facilitated to be carried out later.
Summary of the invention:
A handwritten electric element symbol recognition method based on singular value decomposition comprises four steps of image preprocessing, element standardization, feature extraction and element classification. The image preprocessing is mainly to preprocess the image through operations such as relevant knowledge of digital image processing, filtering and the like, eliminate redundant information in a researched object and standardize the form and the characteristics of the image; the element standardization is to unify the size, format and other characteristics of the element to be identified according to a certain standard and prepare for the subsequent identification and classification process; the characteristic extraction is to link the handwriting electric element with the standard electric element according to the related thought of singular value decomposition to extract the characteristic, thereby being beneficial to the characteristic identification of the handwriting electric element; and in the element classification process, a nearest neighbor classifier is adopted to classify the samples, and the accuracy of the identification method is judged.
the technical scheme of the invention is as follows:
A handwritten electric element symbol recognition method based on singular value decomposition comprises a training part and a testing part; the training part comprises the following steps:
1-1) carrying out binarization and denoising treatment on the electric power engineering drawing;
1-2) cutting and standardizing the elements and the handwritten electric element symbols in the image processed in the step 1-1);
1-3) training a singular value decomposition model by using a singular value decomposition idea;
The test part comprises the following steps:
2-1) cutting and standardizing the handwritten electrical element symbols;
2-2) identifying the element processed in the step 2-1) through a trained singular value decomposition model.
According to a preferred embodiment of the present invention, in the step 1-1), the binarization processing method includes: the method comprises the steps of performing square-evolution operation on the third dimension of a handwriting element image and an electric power engineering image to reduce a three-dimensional matrix to a two-dimensional gray image matrix, and then converting the two-dimensional gray image matrix into a matrix with only 0 and 1 by selecting a proper threshold (for a person skilled in the art, the selection of the threshold is generally experience adjustment and can be realized according to the prior art); for the electric element, the color information and the gray information belong to redundant information in identification, and meanwhile, a scanned graph of the handwriting electric element has certain redundant information due to the relation of scanning environments, so that the handwriting element image and the electric power engineering image are subjected to binarization processing in a preprocessing stage;
The denoising processing method comprises the following steps: and filtering the binarized matrix through Gaussian filtering and mean filtering for resisting pepper noise, thereby effectively removing noise interference. Due to the influence of the scanning environment of the original image and the selected threshold value during binarization, noise inevitably occurs in the binarization matrix, namely some wrong pixel points are introduced, so that the denoising processing is introduced, and the image preprocessing process is finished.
Preferably, according to the present invention, in step 1-1), the method further comprises: and carrying out inversion operation on the matrix after binarization. Because the sensitivity of human eyes to white pixel points is higher than that of black pixel points, the matrix after binarization is subjected to inversion operation, so that element information can be observed by human eyes more easily.
preferably, in step 1-2), the method for cutting and standardizing includes selecting the gravity centers of the handwriting component and the electrical engineering component as the center of the standardized image, correspondingly enlarging and reducing the original image, and standardizing the graphic specification to 100 × 100, that is, each component is guaranteed to be square, the side length is 100 pixels, and the total number of 10000 pixels is total. The purpose of the element standardization process is to match the size and specification of the image of the handwriting element and the size and specification of the element image of the power engineering diagram, so as to prepare for the subsequent singular value decomposition and achieve the standardization effect. The center of gravity as described herein is the geometric center of the element.
preferably, the method for training the singular value decomposition model by using the concept of singular value decomposition in the step 1-3) comprises the following steps:
Our goal is to train a singular value decomposition model, so, assuming that the handwritten electrical element is a probe sample and the electrical engineering drawing element is a galery sample, during the model training process, assume Xm、Ymis a feature of dimension m, wherein Xmfor the original characteristics of the elements of the power scheme, YmOriginal characteristics of the handwriting electric element; the probe sample and the galery sample are different from each other, the singular value decomposition combines the probe sample and the galery sample through a certain means, and the galery sample and the probe sample are combined together to form a training sample so as to achieve the purpose of connecting the two samples; assuming that there are N samples for the galery sample and the probe sample, respectively, the training matrix is assumed as follows:
Wherein each row in the matrix represents a feature and each column represents a number of samples of the feature; the features of the training matrix M, the galery samples and the probe samples are projected to a common subspace;
With the idea of singular value decomposition, the training matrix M will be decomposed into three matrices as follows:
Whereinfor the concatenated features in the common subspace, d represents the feature dimension of each sample,is a diagonal matrix, representing the singular values of the M matrix,The orthonormal basis of the subspace is represented.
U includes two different features, which are independent of each other, and is decomposed as follows:
wherein R isxAnd Ryrespectively representing galery and probe sample characteristic sub-matrixes, namely an electric power process diagram element characteristic sub-matrix and a hand-written electric element characteristic sub-matrix;
suppose x represents the galery feature, i.e., the electric power diagram element feature, and y represents the probe feature, i.e., the hand-written electric element feature:
let K be SVT，kmis the m-th row vector of K, then a point λ on the subspace is represented as
then the electrical engineering drawing element characteristics will be projected to the space of the hand-written electrical element characteristics through point lambda, as follows,
And after the training of the singular value decomposition model is finished, two different features are projected into a common space, so that a good basis is provided for subsequent classification.
According to the invention, the method for cutting and standardizing the hand-written electric element symbols in the step 2-1) comprises the following steps: selecting the gravity center of a handwriting element as the center of a standardized image, correspondingly amplifying and reducing the original image, and standardizing the pattern specification to 100 x 100, namely ensuring that each element pattern is square, the side length is 100 pixels, and the total number of 10000 pixel points is total. The center of gravity as described herein is the geometric center of the element.
According to the present invention, preferably, the method for identifying the element processed in step 2-1) in step 2-2) through the trained singular value decomposition model is as follows:
given a galery sample seti.e. a sample set of electrical engineering drawing element features, passing through sample xicalculating a point λ to obtain a pointThe handwritten electrical component characteristic is the probe sample characteristicis classified as follows by the following categories,
πiwherein
wherein piiis a class label for the sample.
The invention has the beneficial effects that:
a handwritten electric element symbol recognition method based on singular value decomposition comprises training and testing, wherein the training comprises the steps of carrying out binarization and denoising processing on an electric power engineering drawing, and cutting and standardizing elements and handwritten electric element symbols; training a singular value decomposition model by using a singular value decomposition idea; the testing includes cutting and standardizing the hand-written electrical element symbols; and identifying the elements through a trained singular value decomposition model. The image preprocessing is mainly to preprocess the image through operations such as relevant knowledge of digital image processing, filtering and the like, eliminate redundant information in a researched object and standardize the form and the characteristics of the image; the element standardization is to unify the size, format and other characteristics of the element to be identified according to a certain standard and prepare for the subsequent identification and classification process; the characteristic extraction is to link the handwriting electric element with the standard electric element according to the related thought of singular value decomposition to extract the characteristic, thereby being beneficial to the characteristic identification of the handwriting electric element; and in the element classification process, a nearest neighbor classifier is adopted to classify the samples, and the accuracy of the identification method is judged. The method provided by the invention detects and identifies the handwritten electric element by using the thought of singular value decomposition through methods such as machine learning, mode identification and the like, is favorable for effectively utilizing the handwritten electric power engineering drawing, improves the automation efficiency of the electric power engineering drawing, and is favorable for developing related work later.
drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic diagram of the power process described in the present invention;
FIG. 3 is a schematic diagram of a hand-written electrical component.
Detailed Description
the invention is described in detail below with reference to the following examples and the accompanying drawings of the specification, but is not limited thereto.
as shown in fig. 1-3.
Examples of the following,
A handwritten electric element symbol recognition method based on singular value decomposition comprises a training part and a testing part; the training part comprises the following steps:
1-1) carrying out binarization and denoising treatment on the electric power engineering drawing;
in the step 1-1), the binarization processing method comprises the following steps: performing square-evolution operation on the third dimension of the handwriting element image and the power engineering image to reduce the three-dimensional matrix to a two-dimensional gray image matrix, and converting the two-dimensional gray image matrix into a matrix with only 0 and 1;
The denoising processing method comprises the following steps: filtering the binarized matrix through Gaussian filtering and mean filtering for resisting pepper noise, and effectively removing noise interference;
In step 1-1), the method further comprises: carrying out inversion operation on the binarized matrix;
1-2) cutting and standardizing the elements and the handwritten electric element symbols in the image processed in the step 1-1);
in the step 1-2), the cutting and standardizing method comprises the steps of selecting the gravity centers of the handwriting element and the electrical engineering element as the centers of standardized images, correspondingly amplifying and reducing the original images, and standardizing the graphic specification to 100 x 100, namely ensuring that each element is square in graphic, the side length is 100 pixels, and the total number of 10000 pixel points is total;
1-3) training a singular value decomposition model by using a singular value decomposition idea;
in step 1-3), the specific method for training the singular value decomposition model by using the singular value decomposition idea comprises the following steps:
During model training, assume Xm、Ymis a feature of dimension m, wherein XmFor the original characteristics of the elements of the power scheme, YmOriginal characteristics of the handwriting electric element; assuming that there are N samples for the galery sample and the probe sample, respectively, the training matrix is assumed as follows:
wherein each row in the matrix represents a feature and each column represents a number of samples of the feature; the features of the training matrix M, the galery samples and the probe samples are projected to a common subspace;
With the idea of singular value decomposition, the training matrix M will be decomposed into three matrices as follows:
Whereinfor the concatenated features in the common subspace, d represents the feature dimension of each sample,Is a diagonal matrix, representing the singular values of the M matrix,the orthonormal basis of the subspace is represented.
u includes two different features, which are independent of each other, and is decomposed as follows:
wherein R isxand Ryrespectively representing galery and probe sample characteristic sub-matrixes, namely an electric power process diagram element characteristic sub-matrix and a hand-written electric element characteristic sub-matrix;
suppose x represents the galery feature, i.e., the electric power diagram element feature, and y represents the probe feature, i.e., the hand-written electric element feature:
Let K be SVT，kmIs the m-th row vector of K, then a point λ on the subspace is represented as
Then the electrical engineering drawing element characteristics will be projected to the space of the hand-written electrical element characteristics through point lambda, as follows,
The test part comprises the following steps:
2-1) cutting and standardizing the handwritten electrical element symbols;
In the step 2-1), the method for cutting and standardizing the hand-written electric element symbol comprises the following steps: selecting the gravity center of a handwriting element as the center of a standardized image, correspondingly amplifying and reducing the original image, and standardizing the pattern specification to 100 x 100, namely ensuring that each element pattern is square, the side length is 100 pixels, and the total number of 10000 pixel points is obtained;
2-2) identifying the element processed in the step 2-1) through a trained singular value decomposition model;
In the step 2-2), the method for identifying the element processed in the step 2-1) through the trained singular value decomposition model is as follows:
Given a galery sample setI.e. a sample set of electrical engineering drawing element features, passing through sample xiCalculating a point λ to obtain a pointThe handwritten electrical component characteristic is the probe sample characteristicis classified as follows by the following categories,
πiWherein
whereinπiIs a class label for the sample.
Claims (3)
1. a handwritten electric element symbol recognition method based on singular value decomposition is characterized by comprising a training part and a testing part; the training part comprises the following steps:
1-1) carrying out binarization and denoising treatment on the electric power engineering drawing;
1-2) cutting and standardizing the elements and the handwritten electric element symbols in the image processed in the step 1-1);
1-3) training a singular value decomposition model by using a singular value decomposition idea;
The test part comprises the following steps:
2-1) cutting and standardizing the handwritten electrical element symbols;
2-2) identifying the element processed in the step 2-1) through a trained singular value decomposition model;
in the step 1-1), the binarization processing method comprises the following steps: performing square-evolution operation on the third dimension of the handwriting element image and the power engineering image to reduce the three-dimensional matrix to a two-dimensional gray image matrix, and converting the two-dimensional gray image matrix into a matrix with only 0 and 1;
the denoising processing method comprises the following steps: filtering the binarized matrix through Gaussian filtering and mean filtering for resisting pepper noise, and effectively removing noise interference;
in step 1-1), the method further comprises: carrying out inversion operation on the binarized matrix;
In the step 1-2), the cutting and standardizing method comprises the steps of selecting the gravity centers of the handwriting element and the electrical engineering element as the centers of standardized images, correspondingly amplifying and reducing the original images, and standardizing the graphic specification to 100 x 100, namely ensuring that each element is square in graphic, the side length is 100 pixels, and the total number of 10000 pixel points is total;
the specific method for training the singular value decomposition model by using the singular value decomposition idea in the step 1-3) comprises the following steps:
during model training, assume Xm、Ymis a feature of dimension m, wherein XmFor the original characteristics of the elements of the power scheme, Ymoriginal characteristics of the handwriting electric element; assuming that there are N samples for the galery sample and the probe sample, respectively, the training matrix is assumed as follows:
Wherein each row in the matrix represents a feature and each column represents a number of samples of the feature; the features of the training matrix M, the galery samples and the probe samples are projected to a common subspace;
with the idea of singular value decomposition, the training matrix M will be decomposed into three matrices as follows:
whereinFor the concatenated features in the common subspace, d represents the feature dimension of each sample,is a diagonal matrix, representing the singular values of the M matrix,The orthonormal basis of the subspace is represented;
u includes two different features, which are independent of each other, and is decomposed as follows:
Wherein R isxAnd Ryrespectively representing the sublattices of characteristics of the galery and probe samples, i.e. the power worker respectivelya program element characteristic sub-matrix and a handwriting electric element characteristic sub-matrix;
Suppose x represents the galery feature, i.e., the electric power diagram element feature, and y represents the probe feature, i.e., the hand-written electric element feature:
Let K be SVT，kmis the m-th row vector of K, then a point λ on the subspace is represented as
then the electrical engineering drawing element characteristics will be projected to the space of the hand-written electrical element characteristics through point lambda, as follows,
2. The singular value decomposition-based handwritten electrical component symbol recognition method according to claim 1, wherein said method for cutting and normalizing handwritten electrical component symbols in step 2-1) comprises: selecting the gravity center of a handwriting element as the center of a standardized image, correspondingly amplifying and reducing the original image, and standardizing the pattern specification to 100 x 100, namely ensuring that each element pattern is square, the side length is 100 pixels, and the total number of 10000 pixel points is total.
3. the method for recognizing handwritten electric element symbols based on singular value decomposition as claimed in claim 2, wherein said step 2-2) is to recognize the element processed in step 2-1) through a trained singular value decomposition model as follows:
given a galery sample seti.e. a sample set of electrical engineering drawing element features, passing through sample xiCalculating a point λ to obtain a pointThe handwritten electrical component characteristic is the probe sample characteristicis classified as follows by the following categories,
πiWherein
Wherein piiIs a class label for the sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611257061.1A CN106650696B (en) | 2016-12-30 | 2016-12-30 | method for identifying handwritten electrical element symbol based on singular value decomposition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611257061.1A CN106650696B (en) | 2016-12-30 | 2016-12-30 | method for identifying handwritten electrical element symbol based on singular value decomposition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106650696A CN106650696A (en) | 2017-05-10 |
| CN106650696B true CN106650696B (en) | 2019-12-10 |
Family
ID=58837568
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611257061.1A Expired - Fee Related CN106650696B (en) | 2016-12-30 | 2016-12-30 | method for identifying handwritten electrical element symbol based on singular value decomposition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106650696B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106971180B (en) * | 2017-05-16 | 2019-05-07 | 山东大学 | A kind of micro- expression recognition method based on the sparse transfer learning of voice dictionary |
| CN107330434A (en) * | 2017-06-22 | 2017-11-07 | 电子科技大学 | Electrical symbol recognition methods in a kind of circuit diagram based on PHOG features |
| CN107423717B (en) * | 2017-08-01 | 2020-01-14 | 山东大学 | Matching identification method for manually drawn electrical symbols and standard electrical symbols based on joint non-negative matrix binary decomposition |
| CN109117764A (en) * | 2018-07-29 | 2019-01-01 | 国网上海市电力公司 | Using the method for color threshold method identification target object region electrical symbol in power monitoring |
| CN110108474B (en) * | 2019-06-04 | 2020-05-15 | 山东大学 | Online monitoring and evaluating method and system for operation stability of rotary machine |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101833751A (en) * | 2010-04-20 | 2010-09-15 | 南京航空航天大学 | Method for removing thermal blooming fuzzy based on singular value decomposition |
| CN101950422A (en) * | 2010-09-29 | 2011-01-19 | 北京航空航天大学 | Singular value decomposition(SVD)-based image quality evaluation method |
| CN104573729A (en) * | 2015-01-23 | 2015-04-29 | 东南大学 | Image classification method based on kernel principal component analysis network |
| CN105608690A (en) * | 2015-12-05 | 2016-05-25 | 陕西师范大学 | Graph theory and semi supervised learning combination-based image segmentation method |
| CN105868760A (en) * | 2016-03-11 | 2016-08-17 | 信阳农林学院 | Pattern recognition method and system |
-
2016
- 2016-12-30 CN CN201611257061.1A patent/CN106650696B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101833751A (en) * | 2010-04-20 | 2010-09-15 | 南京航空航天大学 | Method for removing thermal blooming fuzzy based on singular value decomposition |
| CN101950422A (en) * | 2010-09-29 | 2011-01-19 | 北京航空航天大学 | Singular value decomposition(SVD)-based image quality evaluation method |
| CN104573729A (en) * | 2015-01-23 | 2015-04-29 | 东南大学 | Image classification method based on kernel principal component analysis network |
| CN105608690A (en) * | 2015-12-05 | 2016-05-25 | 陕西师范大学 | Graph theory and semi supervised learning combination-based image segmentation method |
| CN105868760A (en) * | 2016-03-11 | 2016-08-17 | 信阳农林学院 | Pattern recognition method and system |
Non-Patent Citations (1)
| Title |
|---|
| 基于矩阵SVD的手写数字分类及其特征区域的SVD优化;曹立;《软件》;20160731;第37卷(第7期);第31-37页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106650696A (en) | 2017-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106650696B (en) | method for identifying handwritten electrical element symbol based on singular value decomposition | |
| Afroge et al. | Optical character recognition using back propagation neural network | |
| Bristow et al. | Why do linear SVMs trained on HOG features perform so well? | |
| Cai et al. | Traffic sign recognition algorithm based on shape signature and dual-tree complex wavelet transform | |
| Faraki et al. | Approximate infinite-dimensional region covariance descriptors for image classification | |
| CN112001362A (en) | Image analysis method, image analysis device and image analysis system | |
| Halbe et al. | A novel approach to HTML page creation using neural network | |
| CN105550712A (en) | Optimized convolution automatic encoding network-based auroral image sorting method | |
| CN108133211B (en) | Power distribution cabinet detection method based on mobile terminal visual image | |
| Lahiani et al. | Hand pose estimation system based on Viola-Jones algorithm for android devices | |
| CN115797970B (en) | Dense pedestrian target detection method and system based on YOLOv5 model | |
| Silva et al. | POEM-based facial expression recognition, a new approach | |
| Pauly et al. | Hand written digit recognition system for South Indian languages using artificial neural networks | |
| Nasiri et al. | A new binarization method for high accuracy handwritten digit recognition of slabs in steel companies | |
| CN112818779B (en) | Human behavior recognition method based on feature optimization and multiple feature fusion | |
| Kamranian et al. | A novel free format Persian/Arabic handwritten zip code recognition system | |
| Hachad et al. | Student's attendance management using deep facial recognition | |
| CN112418085A (en) | Facial expression recognition method under partial shielding working condition | |
| Singla et al. | Offline handwritten Devanagari numerals recognition using GLCM features & neural networks | |
| Yongchun et al. | Research of license plate character features extraction and recognition | |
| Jaafar | Survey on currency classification system | |
| CN113807392B (en) | Tampered image identification method based on multi-preprocessing-feature fusion | |
| Fang et al. | Structural sparse representation for object detection | |
| Druki et al. | Development of sequential optimizational algorithms for object detection in images | |
| Swapna et al. | A survey on local patterns for signature verification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20191210 Termination date: 20211230 |