68 results about "Font recognition" patented technology

The embodiment of the invention discloses a handwriting font recognition method, comprising making a pre-acquisition of the user-defined handwriting font and establishing the user-defined handwriting font and a corresponding standard font mapping table; receiving the user input handwriting font and reading the standard font corresponding to the handwriting font. The invention also discloses a handwriting font recognition equipment, capable of fast recognizing a cursive hand connected within words and user-defined words, increasing the input speed of the handwriting font and recognition accuracy.

The invention provides an off-line handwritten Chinese character font recognition method based on a deep convolutional neural network. The off-line handwritten Chinese character font recognition method includes the following steps: 1) collecting and reading Chinese character images; 2) processing the images, including the following steps: 2.1) segmenting the Chinese characters, segmenting all theChinese characters in the images into one Chinese character for one image; and 2.2) compressing or zooming the images, performing normalization on the images as the segmented images cannot satisfy input of recognition, wherein the unified size is 64*64 pixel; and 3) reading the processed image group, performing recognition through the trained deep convolutional neural network, and outputting the result. The off-line handwritten Chinese character font recognition method based on a deep convolutional neural network can improve the recognition effect.

The invention provides a verification method and device based on handwritten character recognition. The method comprises the steps that S1 a reference character is extracted from a candidate character library through a random algorithm, and a picture is generated as the reference verification information; S2 a handwritten character in a handwritten area is acquired, and a picture is generated as handwritten verification information; S3 a font recognition algorithm is used to determine whether the handwritten character of the handwritten verification information matches the reference character of the reference verification information, and if so, the step S4 is carried out, otherwise the verification fails; S4 the feature of the handwritten verification information is extracted to acquire the corresponding eigenvector; and S5 whether the eigenvector of the handwritten verification information matches an eigenvector in a template library, and if not, the verification fails. According to the invention, the handwritten alphabetic character of a designated area is acquired to distinguish human and computer; affects caused by the fact that a verification code is cracked by a malicious program are prevented; and the Internet safety is improved for a user.

The invention discloses an ancient character and font recognition method based on improved YOLO v3. The method is a target detection method based on deep learning. A deep neural network structure is utilized to learn an overall-partial decomposition relationship in an ancient character image, useful feature information is obtained through a feature extraction network, detection and positioning areperformed, then the feature information of the image is sent to a classifier for classification and recognition, and the position of the ancient character is framed in the image by using a bounding box. According to the method, the problems that ancient characters have complex internal structures and the precision is low when the characteristics are used for recognition are solved. According to the improved YOLO v3 provided by the invention, the ShuffleNet v2 is used as a main structure of the model, so that the improved YOLO v3 is more efficient. The accuracy of ancient character and font recognition reaches 98.81%, and the method has good stability and good robustness and can be applied to ancient character recognition scenes such as ancient character texts and inscription stickers.

The invented system suitable for inputting single Chinese character comprises first input unit, second unit, speech recognition unit, font recognition unit, font library unit, Chinese character features processing unit and display unit. The method includes following procedures. The phonetic feature of Chinese characters is input by first input unit. The second input unit inputs N font features ofsingle character. The speech recognition unit and font recognition unit recognize speech feature and font feature. The Chinese character features processing unit combining the speech feature and fontfeature selects out a set of candidate characters, which accord with speech feature as well as accord with N font features, from font the library then outputs and displays the set.

The invention provides a device for identifying stamped characters of a doorplate based on caffe and soft trigger. The device comprises a hardware collection equipment module and a software processingprogram module, wherein the hardware collection equipment module is used for obtaining an image of the stamped character; a camera collection module is used for obtaining and collecting the image, the type is distinguished, and the product information is improved; a shooting module is triggered by software to collect the image; the software processing program module is used for processing the obtained image, splitting the characters and identifying. The device adopts a method with the following steps of firstly, detecting whether the image contains a circle shape with specified size or not, acquiring the satisfactory image, and avoiding overlow definition degree of the collected image; providing a floodfill algorithm to determine the positions of the characters, and eliminate the interference background; finally, calling caffe and LeNet-5 handwriting font identifying networks to learn and identify. Compared with the traditional feature extracting algorithm, the accuracy is obviously improved.

The invention discloses a handwritten font recognition method and system and a terminal device, wherein the method comprises the following steps of collecting handwritten font tracks when a user writes, and obtaining the corresponding standard font characters according to the handwritten font tracks; carrying out feature extraction of handwritten font trajectory to obtain feature information of handwritten font text, and creating a text library according to the feature information of handwritten font text and corresponding standard font text; collecting an image of a handwritten manuscript andprocessing the image of the handwritten manuscript to obtain characteristic information of each handwritten font character in the handwritten manuscript; selecting a mapping relation set of a user corresponding to an author of a handwritten manuscript in a text library and matching the mapping relation set by calling the selected mapping relation set to obtain a standard font character corresponding to each handwritten font character in the handwritten manuscript; generating a standard font manuscript corresponding to the handwritten manuscript according to the standard font manuscript. The invention can quickly recognize the handwritten fonts of handwritten manuscripts, and the recognition accuracy is improved.

The invention discloses a handwritten font recognition method, a system and a terminal device, wherein, the method comprises the following steps: obtaining handwritten font characters written by a user, extracting features of the handwritten font characters to obtain feature information of the handwritten font characters, and creating a text library according to the feature information of the handwritten font characters; collecting an image of a handwritten manuscript and processing the image of the handwritten manuscript to obtain characteristic information of each handwritten font character;Selecting a mapping relation set of the user corresponding to the author of the handwritten manuscript in the text library, and matching the feature information of each handwritten font character inthe handwritten manuscript by calling the selected mapping relation set to obtain a standard font character corresponding to each handwritten font character; a standard font script corresponding to the handwritten manuscript is generated according to the standard font script corresponding to each handwritten font script. The invention can quickly recognize handwritten fonts of handwritten manuscripts, and the recognition accuracy is greatly improved.

The invention provides a ship water gauge marking identification method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining a ship water gauge image of atarget ship; identifying water gauge scale data in the ship water gauge image according to a pre-trained font identification library; carrying out edge detection on the ship water gauge image to obtain a plurality of straight lines; determining a water line according to the plurality of straight lines and the water gauge scale data; and determining the reading of the water gauge marked line according to the water gauge scale data and the water level line. According to the method, automatic identification of the ship water gauge marked line is realized, the identification efficiency of the water gauge marked line can be greatly improved, and the labor cost is saved.

The invention provides a character recognizing method based on a sparse coding. The method provided by the invention takes a gray level image as an input to carry out operations of following two phases on any one image to be tested: in a training phase, the image to be tested of each character is randomly divided into a certain number of diamonds and the number of the diamonds is codetermined by the size of the image and the size of the diamond: for example, a gray level image with the size of 512*512 can be divided into 4096 diamonds with the size of 8*8. To any one type of character, the divided diamonds are used as input and a group of primary function capable of sparsely representing any diamond is trained through utilizing an independent component analyzing method and the primary functions are used as models of the characters. The method provided by the invention can recognize Chinese characters, can recognize characters of other languages and also can recognize characters of different languages. The method can be applied to automatic document analysis, article design and the like.

The embodiment of the invention provides a font recognition method and a device. The method comprises the following steps: obtaining a picture containing a character to be recognized; Segmentation ofthe picture to obtain an object to be identified in the segmentation region; The target to be recognized comprises at least one character to be recognized; According to the target to be recognized andthe pre-trained convolution neural network CNN model, the font of the character to be recognized is recognized. The apparatus performs the method described above. The font recognition method and thedevice provided by the embodiment of the invention can recognize the font efficiently and accurately by acquiring an object to be recognized in a segmentation area of a picture, wherein the object tobe recognized comprises at least one character to be recognized, and then identifying the font of the character to be recognized through a CNN model.

The invention provides a homework correcting method and device, equipment and a storage medium. The homework correcting method comprises the steps: collecting pictures of homework to be corrected through terminal equipment; performing question-by-question segmentation on the picture through feature point identification, and dividing the picture into a plurality of sub-pictures; segmenting the sub-picture into a question part and an answer part based on font recognition, and associating the question part with the answer part; when the sub-pictures are corrected, employing the question parts ofthe sub-pictures for retrieval in a preset picture library to obtain standard answers corresponding to the question parts of the sub-pictures; obtaining a question type to which the question part of the sub-picture belongs, and loading a corresponding correction method based on the question type; and completing correction of the title part of the sub-picture through a correction method. Accordingto the homework correcting method, the corresponding correcting method can be loaded based on the question type, the question type capable of being corrected is expanded, and the homework correcting efficiency is improved.

The invention discloses a system and method for recognizing handwritten fonts. The system comprises a flexible wearable bracelet, a sensor signal obtaining and processing unit, and a convolutional neural network unit, and the flexible wearable bracelet is electrically connected with the sensor signal obtaining unit and the convolutional neural network unit. Handwritten font recognition based on the flexible wearable wristband eliminates the problems of ambient light and resolution, and is more stable and lower in cost; by means of the convolutional neural network, the handwritten fonts are recognized through the trained model, the time consumption is short, the recognition rate is high, and the robustness is higher; according to the method, the output layer structure can be flexibly modified, and the method can be suitable for various specific occasions through transfer learning.

The invention relates to the technical field of monitoring, and discloses a printed file anti-forgetting method and system, electronic equipment and a storage medium, and the method comprises the steps: collecting a printed file image every other preset time; according to a preset special font recognition model, performing image recognition on the currently acquired printed file image to obtain aspecial font type in the current printed file image or position information of characters with the special font type; obtaining a current feature code according to the conversion rule and the specialfont type in the current printed file image or according to the conversion rule and the position information of the characters with the special font type in the current printed file image; judging whether the previous feature code is the same as the current feature code or not; and if yes, querying the employee ID corresponding to the feature code of the currently printed file image in the featurecode library, and sending reminding information to the terminal corresponding to the employee ID. The technical problem of information leakage risk caused by forgetting of a printed paper document inthe prior art is solved.

The invention provides a homework correction method and device and electronic equipment, and the method comprises the steps: detecting a text line of a homework image through a neural network model after the homework image uploaded by a student is obtained, and determining a printed character text line and a handwritten character text line of the homework image; respectively inputting the homework image of which the text line is determined into a first font recognition model and a second font recognition model, recognizing a printed character text line and a handwritten character text line in the homework image, and determining the recognized printed character text line in the homework image as a homework question; and determining the recognized handwritten character text line in the homework image as the answer to the homework question of the student, so that the homework uploaded by the student is corrected automatically, a teacher does not need to correct a large amount of homework personally, and a large amount of homework correction time is saved.

The invention provides a multi-style font generation method based on StarGAN improvement, original generator network structures of a coding layer, a conversion layer and a decoding layer are replaced by combination of two special structures of Attention U-NET and DenseNet, and the problem that character stroke information is lost due to direct convolution operation is solved. Meanwhile, a U-shaped structure and jump connection in the Attention U-NET can effectively integrate details and semantic features of the image, the attention mechanism of the network can solve the problem that local information is lost or redundant in the font generation process of an existing model, and the pixel error rate of Chinese character generation is reduced. According to the method, a paired data set is not needed, and automatic generation of various font styles can be directly realized only by training one model. The method solves the problem of data set missing in Chinese character recognition, improves the font recognition accuracy, and creates conditions for insufficient quantity in the small sample field and the data migration field.

The invention provides a method and a system for training a sensitive word detection model. The method comprises the steps: step A-1, inputting sample data of a training corpus into a first BLSTM model and a second BLSTM model, inputting outputs of the first BLSTM model and the second BLSTM model into a CRF model, and outputting a sensitive word recognition result of an input text by the CRF model; updating the current parameters of the model based on the difference between the identification result of the CRF and the marking result of the input text; step A-2, inputting the sample data of thetraining corpus into a current first BLSTM model, inputting the output of the first BLSTM model into a CNN model, and outputting a font recognition result of an input text by the CNN model; and updating the current parameter of the model based on the font difference between the recognition result of the CNN and the input text. According to the method and system for training the sensitive word detection model, the sensitive word detection model with better performance can be obtained, and compared with a traditional DFA algorithm, the sensitive word detection is not limited by a sensitive wordlexicon and has a certain detection capability on foreign characters.

The invention discloses a font recognition method, which comprises the following steps of: character recognition: obtaining a to-be-retrieved text, sequentially generating corresponding quadrangular codes from text characters according to a quadrangular code coding principle, and correspondingly searching a font library to obtain stroke number information of each character of the to-be-retrieved text; and a comparison step: analyzing and comparing the four-corner code data strings arranged in sequence with a comparison database to generate a font similarity list, and judging whether a font similarity condition exists or not according to the font similarity list in combination with a stroke number information comparison result of each word of the to-be-retrieved text. The invention furtherdiscloses electronic device and a storage medium, the four-corner codes of the characters are obtained from the text characters to be retrieved, the stroke number information corresponding to the characters is combined and compared with the comparison database, and the font similarity condition is judged by integrating the two comparison results. And an image recognition and comparison method is abandoned, and image recognition errors are eliminated, so that the font similarity judgment is more accurate.

The invention discloses an OCR identification method based on localization platform and identification system thereof, and belongs to the technical field of pattern recognition and computer vision. The method comprises the following steps: arranging characters to be recognized, making a data set according to the collected characters, and performing data augmentation processing; establishing a deep learning framework based on an image processing tool of a localization platform, and establishing an OCR network structure; performing image size normalization on the to-be-trained data by using a compiled image processing tool, setting training parameters for training, and performing iterative optimization on the parameters of the network structure in the training process; and finally, sending the processed picture into the trained network structure to calculate a most matched label result, and outputting an OCR identification result. According to the method, confidentiality and autonomous controllability of the technology are achieved, the character recognition effect is good, and high robustness is achieved for bankbook, invoice and deposit receipt type font recognition.

The present invention relates to the field of computer technology, in particular to the field of pattern recognition and depth learning, and more particularly to an OCR font-based similar character recognition method. By changing the traditional font recognition method, both character text and font can be recognized. By comparing multiple samples and adding threshold filter, not only the accuracyof text recognition is greatly improved, but also the character font can be recognized effectively. The method is especially suitable for similar font and similar font character recognition, to achieve the double accurate recognition of font and font. Each character is cut into 96*96 pixels in size by horizontal and vertical segmentation, the method facilitates the extraction of pixel characteristic information, avoids the mutual interference between adjacent characters, and effectively improves the recognition efficiency. The designer of the invention cuts each character into 96*96 pixels invarious pictures such as books, newspapers, clothes and screen shots to extract the character pixel characteristic information, and the extraction rate is close to 100%.

The invention relates to a font recognition method based on image processing, belonging to the technical field of font recognition. As that TTC font file are collected, from which character feature vectors are extracted to form a font feature vector database, the image form information of a single character and the corresponding character text information are uploaded by a user, by using modern cutting technology and character shape features, graying and binarization are performed on the image specifically and then, the character features are extracted and the feature vectors are generated. The font recognition degree is calculated with the data of all characters input by users in the database, and the font set with the highest recognition degree is outputted through sorting. Compared withthe prior art, the present invention mainly solves the problems of poor accuracy and inflexible human-computer interaction when the prior art recognizes the font, and increases the accuracy and flexibility of the current font recognition relying on the computer.

The embodiment of the invention provides a font recognition method and device, and the method comprises the steps: determining a to-be-recognized image which comprises to-be-recognized characters; andinputting the to-be-identified picture into the font identification network model to obtain the font of the to-be-identified character in the to-be-identified picture, wherein the font recognition network model is a model obtained by training according to a training set, and the training set comprises a plurality of sample pictures including a character and a real font of the character in each sample picture. As the scanned copy can be converted into the format of the picture, the font identification method provided by the embodiment of the invention can be used for identifying the font of the character in the scanned copy.

The invention discloses a printing font recognition method, a device and a storage medium. The method comprises steps: scanning a printout to be identified; converting a print to be identified into anelectronic picture; determining a target text in the electronic picture; cutting the target text according to a preset size to obtain a plurality of target fonts; a plurality of target fonts and a reference font being multidimensionally parsed and a parsing result being obtained; outputting a result. Embodiments of the present application also provide corresponding apparatuses and storage media.The method provided by the embodiments of the present application can significantly improve the recognition rate of printed fonts.

The invention provides a dictation interaction method, system and device based on AI vision. The method comprises the steps of: S100, obtaining a collected target image in real time; S200: constructing and training a plurality of convolutional deep neural networks and cyclic deep neural networks, or based on a Transformer deep neural network combined structure of a self-attention mechanism, carrying out comprehensive weighting calculation on a plurality of combined structure output results for handwritten font recognition by utilizing a dynamically planned common substring matching algorithm,and recognizing action information and text information in the target image; S300, according to the identified action information, executing a dictation control task or a dictation control task; S400,controlling to play dictation content of the dictation task; and S500, controlling and displaying prompt content and a dictation result in the dictation task. Through multiple convolutional deep neural networks, interaction between gestures and dictation equipment is realized, the recognition accuracy is improved, the recognition speed is increased, and the use experience of a user is enhanced.

The invention provides an international phonetic sign typeface identification method based on feature matching, and relates to the technical field of typeface identification. The method comprises the following steps that international phonetic sign images with the unified dimension are subjected to preprocessing, i.e., each international phonetic sign image is subjected to fining processing, the typeface part in the image is scanned line by line under the condition of ensuring that the continuity of the typeface pixel is not interfered, further, excessive pixel points are removed, and the preprocessed international phonetic sign typeface images are obtained; feature points are extracted; the feature points of each training typeface sample are used for building a template base; input typeface images are tested, feature points are matched, and test results are obtained. The method provided by the invention is utilized, the correct identification on the international phonetic sign typeface can be carried out through the feature point matching, simplicity and high efficiency are realized, in addition, the method can also be used in the typeface identification of other languages, and the universality is good.

The invention provides a hand-written font recognition method based on two-dimensional convolution dimension reduction, for solving the problem that in some complicated situations, such as high deformation, a current handwritten recognition algorithm is not well in high-dimensional data effect. The hand-written font recognition method based on two-dimensional convolution dimension reduction includes the steps: obtaining a special convolutional neural network by adding an optimization layer, so as to enable the network to be able to perform recognition and dimension reduction at the same time;designing a new linear determination analysis target function to reduce the complexity of the optimization process, and lowering the information dimension utilized by the network during the process ofrecognizing complex fonts and simplifying recognition by optimizing the function; and finally recognizing the hand-written font images by means of the trained network, to obtain the recognition result. As the hand-written font recognition method based on two-dimensional convolution dimension reduction uses the specially designed convolutional neural network to optimize the new linear determination analysis target function, and can obtain better recognition performance.

The invention relates to an improved two-step linear discriminant analysis method. The method specifically includes the following steps that: an approximate matrix method is adopted to solve the problem of the relatively high operation complexity of a TSLDA algorithm, an approximate eigenvalue matrix is adopted to replace an original eigenvalue matrix, so that the singularities of Sw and Sb are eliminated; and a screening and compression method is adopted to extract feature vectors in a TSLDA projection space which are favorable to classification, and single-feature Fisher information quantityis introduced to select feature vectors that meet conditions, and the selected feature vectors form an optimal projection space, thus, the projection space can be compressed, and the validity of allfeature information in the projection space can be ensured. The improved two-step linear discriminant analysis method has a high application value in fields such as image recognition including face recognition, fingerprint recognition, and handwriting font recognition.

Embodiments of the invention provide a font recognition method and apparatus. The method comprises the steps of obtaining a to-be-identified text picture; selecting at least one sliding window of a preset scale from the plurality of sliding windows of the preset scale; scanning the text picture by adopting the at least one sliding window with the preset scale to obtain a plurality of sub-text pictures; and obtaining the font of the text in the text picture according to the plurality of sub-text pictures and a machine learning model. According to the embodiment of the invention, the fonts of the texts in the text pictures with different text lengths can be identified.

The present invention relates to a font recognition method and system, the method comprising: obtaining a field to be recognized, recognizing Chinese and English in the field to be recognized, respectively obtaining a Chinese field and an English field; calculating characters contained in the Chinese field and the English field respectively Chinese features and English features; match and recognize the Chinese features and English features with the pre-set recognition model respectively, and obtain the Chinese font and English font of the field to be recognized. The font recognition method and system provided by the present invention realize the recognition of fonts including multiple different language situations, and can improve the speed and accuracy of recognition.