70 results about "Image name" patented technology

Embodiments disclosed herein utilize Habitats to perform incremental updates on virtual machines (VMs) over time. A VM Habitat comprises a collection of VMs and a specification shared between these VMs. Each Habitat specification may define an image name, processing power, etc. A Habitat specification may exist across a cluster of servers. In some embodiments, to dynamically update a Habitat specification, only “spare” VMs in the Habitat are updated, subject to some scheduling parameters and server concurrency limit. As spare VMs are updated, they move towards the new Habitat specification. Eventually, all VMs reach this goal over time, given sufficient spare VM resources. This approach can allow the service provided by the VMs to remain operational, thus maintaining the service level required for its users. This automated process can also eliminate or otherwise reduce human error either in the server removal / insertion from / to the cluster or in the VM update itself.

The invention relates to a deep learning-based rapid tunnel crack identification method, which mainly solves the problem that the current deep learning-based crack detection method cannot directly acquire the length and width information of cracks. The deep learning-based rapid tunnel crack identification method comprises the following steps: (S1) a deep learning image training set is created; (S2) a deep convolutional neural network model is trained; (S3) the trained convolutional neural network model is used for detecting a to-be-detected image and outputting a prediction label image; (S4) adetection result, which includes image category, coordinate information of cracks and pixel-level width values and length values of cracks, is outputted according to the prediction label image; (S5)a disease record result is outputted according to the detection result; if cracks exist in the to-be-detected image, then the image name, the coordinate information of the cracks and the actual widthvalues and length values of the cracks are recorded; and if no cracks exist in the to-be-detected image, then recording is not carried out.

The invention discloses a railway wagon brake shoe key going-out fault recognition method based on an artificial neural network. The method comprises the following steps: (1) obtaining an image, model and wagon information of a running railway wagon body when a train passes through; (2) reading a to-be-recognized image; (3) performing preliminary positioning according to an image name to obtain apreliminarily positioned image; (4) scaling-down the preliminarily positioned image in an equal proportion; (5) preliminarily preprocessing the image scaled down; (6) positioning a candidate region for the preliminarily preprocessed image; (7) extracting an image feature vector of the candidate region; (8) carrying out recognition computing on the feature data; (9) outputting a recognition result. The method cannot influence the normal work of a train inspector on the existing TFDS (travelling fault diagnosis system) system; an automatic recognition result can be stored in a friendly form to provide a convenient re-inspection manner for the train inspector; the method has the advantage of full automatic start or stop and is free from manual intervention in service.

The embodiment of the invention provides a method and apparatus for image retrieval. The image retrieval method comprises the following steps: acquiring a target image to be retrieved, and preprocessing the target image; extracting color features, texture features and image names of target images to generate target feature vectors; inputting the target feature vector into a pre-established image retrieval model, so that the image retrieval model calculates the similarity between the target feature vector and the feature vector of each candidate image in the preset image feature library, determines each candidate image corresponding to the feature vector that meets the similarity judging condition as a desired image, and outputs the determined desired image from the preset image database. When the method provided by the embodiment of the invention is used for image retrieval, the subjectivity problem similar to that of adding keywords in the traditional image retrieval mode by the manual identification mode is not involved, so that the error probability in feature matching can be avoided, and the accuracy of the image retrieval result is greatly improved.

The invention discloses a method and a device for switching the language of mobile terminal game software. The method comprises the following steps: determining the picture resource storage path of a target language by the game software, and reading the picture main body names of pictures recorded in a picture resource configuration file; for each picture main body name, combining the picture main body name with prefix / postfix information of the present language so as to obtain the picture name of a picture to be unloaded, sending out a picture unloading notice carrying the picture name to a game engine, and unloading corresponding pictures by the game engine; then, for each picture main body name, combining the picture main body name with prefix / postfix information of the target language so as to obtain the picture name of a picture to be loaded, sending out a picture resource storage path carrying the target language and a picture loading notice of the picture name to the game engine, and loading corresponding pictures under corresponding paths by the game engine. Therefore, the language switching of pictures containing language and word information is realized at a lower development cost.

The invention involves in mobile communication terminal equipped with internal and external cameras and method for naming stored images thereof. Internal cameras will save the photography and mobile images according to established name by said terminal. If the user wants to modify the names of the photos, he must callout edit function to reedit. It is inconvenient to the user. And it is difficult to manage saved images by giving the same name to all saved images. When taking a picture or shooting using internal cameras, using location service technique, representative placename of its surroundings will be achieved and the name based on it will be added to shot images automatically. It is swift and convenient to manage the images using the name. When said internal memory capacity keeps increasing, the user will get great convenience by using the function provided in the invention of adding image names for said terminal to search and appreciate.

The invention discloses a rapid image retrieval method, relating to a method which retrieves and displays digital images; the method solves the problem of complicated retrieval input. The method of the invention comprises the following steps: the process of creating and storing the comparison table between an image name assembly and a simplified name alphabet on an image processing device; the process of storing images on a storage medium; the process of establishing an image file chain table on the image processing device based on the images stored on the storage medium; the process of creating a hash table of the image simplified names on the image processing device based on the alphabet of the comparison table and the image file chain table; and the process of retrieving the images on the storage medium by simplified names and displaying the thumbnails of the retrieval result on the image processing device through the hash table and the image file chain table. The method of the invention enables the operator to input the simplified names of the images during the retrieval process, so as to avoid the image names expressed in complicated vocabulary; therefore, the method requires small input amount and is simple to operate, which occupies small storage space and has fast speed in the middle process; the method is suitable for hand-held image browsing devices.

The invention discloses an image classification method, a convolutional neural network generation method for image classification processing, a convolutional neural network generation method for recognizing words in images, a mobile terminal and calculation equipment. The image classification method is implemented in the mobile terminal; the mobile terminal comprises an image library; a pluralityof images are stored in the image library; the image classification method comprises the following steps: classifying images of the image library so as to obtain corresponding types; if a type is a text type, carrying out word recognition on an image, and extracting text information of the image; associating and storing the text information with an image storage route and an image name of the image.

The invention provides a processing method and device and a reading method and device for Docker image cache information. The processing method for the Docker image cache information comprises the steps that images in an image warehouse are grouped, and image labels with image names being identifiers and containing all the images, image cache information of cache information of the image labels and image group cache information with the image names being the identifiers and containing image group cache information are generated. Based on the method, by processing the cache information of the images in advance, a basis is provided for directly reading image cache content corresponding to an image cache request from the image cache information and the image group cache information subsequently, in this way, redundant research, traversing and calculation processes are avoided, and the purpose of quickly processing the request is achieved.

The invention provides an electronic device, and an image processing method and system of an application. A first cache mapping table is established for key values according to resource sizes, partial bytes and postfix names of downloaded pictures, and the pictures are cached locally according to the first cache mapping table. When the pictures need to be downloaded subsequently, the matched pictures can be searched locally according to the postfix names, resource sizes and partial bytes of to-be-downloaded pictures; if there are matched pictures, the pictures are directly obtained; and if there is no matched picture, the pictures are downloaded from a control end. The pictures are prevented from being downloaded repeatedly, and the picture obtaining efficiency is high; matching is carried out for the key values according to the postfix names, resource sizes and partial bytes of the pictures, and the matching accuracy is improved; and the problem that in the prior art, the processing efficiency is reduced due to the fact that the matching is carried out for the key values only according to the picture names, the matching result is inaccurate, and the pictures need to be downloaded repeatedly is prevented.

The invention discloses a binary image feature similarity discrimination method based on a random forest algorithm, comprising an off-line index phase: extracting the scale invariant feature transform conversion features of an image, employing each dimension of all features as a vector to be used for clustering through a K mean value method, and quantifying scale invariant feature transform conversion features into 512 dimension binary system features; and writing quantified features, feature indexing, image names corresponding to features and adjacent features into a database to be used as an image feature database. In the off-line index phase, the scale invariant feature transform conversion features of the image are extracted, and are quantified to be 512 dimension binary system features to match features in the image feature database, and find adjacent features; and then a random forest algorithm is employed to discriminate the adjacent features, and a voting mechanism retrieves similar images.

The invention discloses a method for a mobile terminal to browse continuous related pictures through a network. Not all the prior mobile terminals support stream media, video and the like. The methodfor the mobile terminal to browse the continuous related pictures through the network comprises that: a content server stores a group of static picture files in the content server; the mobile terminalcontinuously sends requests to the content server at fixed time intervals; and the content server retrieves picture sequences according to unique continuous signs of picture names, and sends the picture sequences to a client mobile terminal at the fixed time intervals according to the order of the continuous signs so as to continuously display the picture sequences on a display screen of the mobile terminal. The method is used for the mobile terminal to browse video contents in a format of continuous related pictures through the network.

The invention discloses an image stabilization-based remote sensing video satellite image processing method and system, and the method comprises the steps: carrying out the cataloguing processing of each frame, extracting the imaging time of each frame, enabling the imaging time of each frame to serve as a part of an image name, and enabling a plurality of image frames to form an image frame sequence group according to the time sequence; performing relative radiation correction on each frame of the sequence group to eliminate atmospheric radiation; after radiation correction, performing systemgeometric correction on each frame of the sequence group to eliminate geometric distortion generated in the frame during imaging; Taking the previous frame as a reference frame, and carrying out image stabilization processing on each current frame except the first frame in the sequence group, including motion estimation, motion compensation and image correction; and combining the image frame sequences subjected to image stabilization processing to form a video. In the processing process of the remote sensing video satellite, the imaging characteristics of a general remote sensing satellite and a video satellite are considered, and the processing process obviously improves the quality of the remote sensing video.

The invention discloses a super-large image rapid matching splicing method based on block subimage search, and the method is used for speeding up the matching, and improving the matching efficiency. The method comprises the steps: firstly reading and displaying two large images named as A and B through employing a GDAL secondly carrying out the partitioning and cutting of the two large images in a certain mode, obtaining a plurality of subimages, taking one subimage of the image B as the target image, searching a subimage with the maximum similarity from the subimages of the image A, and carrying out the feature matching of the pair of subimages if the similarity meets a condition; selecting another subimage of the image B as the target image if the similarity does not meet the condition, carrying out the search in the image A, and carrying out the loop operation according to the above steps; renaming the images A and B if no subimage meeting the condition is searched, interchanging the names A and B of the images A and B, carrying out the above partitioning and search processes again, and selecting the pair of subimages meeting the condition for feature matching. The method can speed up the matching of the large-size images, and improves the registering efficiency.

The invention discloses a handwritten sample picture marking method, device, a computer device and a storage medium. The method comprises the following steps: obtaining a handwritten sample picture tobe annotated and a text content of handwritten text contained in the handwritten sample picture; Carrying out image preprocessing on the hand-written sample picture to obtain a binary image corresponding to the hand-written sample picture, and obtaining a picture size of the binary image; According to the pixel value of each pixel point in the binary image, determining the corresponding target pixel point of the handwritten text in the binary image, and obtaining the relative position information of the handwritten text in the handwritten sample picture according to the target pixel point; Setting the picture name of the handwritten sample picture; Combining the picture name, the text content, the relative position information and the picture size, the labeling information of the handwritten sample picture to be labeled is obtained. The technical proposal of the invention improves the labeling efficiency of the sample and the integrity and accuracy of the labeling content.

The invention provides a method for searching names related to thematic words from a network. The method comprises the following steps of: searching web pages related to the thematic words in the network and filtering and analyzing the web pages; according to image nodes in the DOM (Document Object Model) tree of each web page, extracting image names and matching the image names with surrounding texts to form a candidate name first set; converting the DOM trees of the web pages into code sequences, determining repeated subsequences in the code sequences, and extracting candidate names corresponding to the repeated subsequences from the candidate name first set of the web pages to form a candidate name second set; filtering names in the candidate name second set according to preset rules and preset templates to determine the candidate names of the web pages; aiming at the candidate names from multiple web pages, filtering the candidate names according to a relationship between the candidate names or between the candidate names and the thematic words to obtain the names related to the thematic words; and calculating the score of each name and ordering the names according to the scores. The invention correspondingly provides equipment for searching the names related to the thematic words from the network.

The embodiment of the invention provides an image uploading method, an image uploading device, an electronic terminal and a readable storage medium. The image uploading method comprises the steps of:starting a network monitoring type object to monitor the network state of the electronic terminal, wherein the network state includes a network type and a network connection state; when the network state is in a target network connection state, providing an operation interface used for receiving configuration information of one or more images to be uploaded according to operation of users, whereinthe configuration information includes image data, image names and image formats; and uploading the images to be uploaded to a server according to the configuration information.

The invention relates to the technical field of artificial intelligence, in particular to a training method and device for an image recognition model, and the method comprises the steps: obtaining to-be-recognized images collected by image collection equipment; for each to-be-identified image, based on a trained coordinate detection model, taking any to-be-identified image as an input parameter, identifying coordinates of a to-be-identified object in the to-be-identified image, and obtaining position information of the to-be-identified object; adding each to-be-identified image, the corresponding position information and a preset object label into a training sample set, the object label being determined based on an image name of a to-be-identified object included in the to-be-identified image; and training an image recognition model based on the training sample set, and obtaining the trained image recognition model, so that the to-be-recognized image is standardized through the trainedcoordinate detection model, the image labeling efficiency can be improved, and the training efficiency of the image recognition model is improved.

The invention provides a picture management method and equipment. The method comprises: obtaining picture information of each to-be-classified picture and commodity information of each to-be-associated commodity, wherein the picture information comprises a picture name, and the commodity information comprises commodity categories and commodity names; obtaining associated commodities correspondingto the to-be-classified pictures according to the picture names of the to-be-classified pictures and the commodity names of the to-be-associated commodities; and obtaining associated commodity categories of associated commodities corresponding to the to-be-classified pictures, determining picture categories of the to-be-classified pictures according to the associated commodity categories corresponding to the to-be-classified pictures, and performing preset management operation on the pictures of which the picture categories are specified picture categories. According to the method, batch management operation can be carried out on the pictures of the same type, batch management of the pictures is achieved, the operation time is shortened, and the picture management efficiency is improved.

The invention discloses a target detection method for a large-size aerial remote sensing image, and belongs to the technical field of target detection. The invention aims at solving the problem that contextual information of a large-size aerial remote sensing image is lost due to image segmentation. The method comprises the following steps: determining an original image name; configuring a subgraph name for a subgraph; retrieving in an original remote sensing image set according to the subgraph name to obtain a corresponding current original remote sensing image; performing feature extraction on a current to-be-detected sub-graph by adopting a local feature extractor to obtain a group of multi-scale sub-graph feature graphs; performing down-sampling on the current original remote sensing image, and performing feature extraction by using a global feature extractor to obtain a group of multi-scale original image feature graphs; adopting a feature pyramid network of a global and local coupling mechanism to obtain a fused feature graph; respectively obtaining a target prediction result and a filtered global prediction result; re-fusing the fusion target prediction result and the filtered global prediction result to obtain a detection target. The invention is used for target detection.

The invention relates to a seasonal style conversion model and method for an image named MSGAN. The seasonal style conversion model comprises a generator G, a generator F, a first true and false discriminator DG, a second true and false discriminator DF and a seasonal discriminator DS. The generator aims to convert an input image into other specific seasonal styles. The true and false discriminator is used for distinguishing whether the image is a composite image. And the season discriminator performs season classification on each synthesized image and each real image. The two discriminators can provide guidance for the generator respectively. In order to give a correct optimization direction of the network, the MSGAN respectively uses style loss, structural similarity loss and color lossto improve the generation capability of the generator. Moreover, the MSGAN uses the saliency information of the image for the first time to guide the image style conversion task, so that the image style conversion result better conforms to the real situation of human eyes.

The invention provides a method and a device for pushing an alarm snapshot picture of terminal equipment, and the method comprises the steps: obtaining a first alarm message which is uploaded when the terminal equipment triggers an alarm event, and the alarm message comprises a unique identifier of the terminal equipment and alarm time; generating a picture name and a picture access path according to the unique identifier of the terminal equipment and the alarm time, and generating an alarm record; pushing a second alarm message to the user terminal, wherein the second alarm message comprises a unique identifier of the terminal equipment, alarm time and a picture access path; acquiring a snapshot picture uploaded by the terminal equipment, wherein the name of the snapshot picture is generated according to the unique identifier of the terminal equipment and the alarm time; and processing the captured picture, and storing the processed captured picture to the corresponding alarm record according to the picture name, thereby solving the problem that in the prior art, the picture uploading is extremely slow, so that a user cannot instantly sense the alarm of the terminal equipment.

The invention relates to a text recognition technology in the field of computer vision, in particular to a bill text recognition method based on a novel network, which comprises the following specificsteps: step 110, acquiring image data of a bill; step 120, creating a text recognition data set; using the text line image data obtained by the text detection module to carry out input and quality inspection on the text line image data, writting a text label into txt or a text line image name, and dividing the data into a training set, a test set and a verification set according to a certain proportion; compared with a common text recognition method, the method is higher in generalization and better in robustness by means of a feature extraction and image correction algorithm in deep learning, can effectively cope with severe conditions such as bill distortion, perspective, printing body ink mark breakage and dirt; by adding BN layers to different levels, the recognition accuracy is higher.

The invention relates to the field of computer big data processing, in particular to a rapid search method for mass images based on a Hadoop platform. The method comprises the steps that 1, a Hadoop cluster platform is built; 2, a security scheme is set; 3, single-image storage processing is performed; 4, file preprocessing and combining are performed; 5, an image index is built; 6, a client side adopts image names and creating time as parameters for initiating an access request, Name Node operation is performed to obtain a minute time period where the images are located and Blocks information corresponding to a combined file, and the minute time period and the Blocks information are returned back to the client side. The problem that Name Node memory consumption is excessive and the search efficiency is low during Hadoop search of mass images can be well solved, the Name Node load during search is effectively reduced, the Name Node performance is improved, and therefore wider application range of the Hadoop platform is achieved.

The invention provides a method for establishing an analysis report. The method comprises the steps that a template document of the analysis report is established according to demands; labels are added in the template document of the analysis report, wherein the labels can be classified into grouping labels, changeable text description labels, picture name labels, table labels and sub-template labels; original data forms content data of the analysis report according to designated format; according to data corresponding to each label name in the content data of the report as well as correspondence relations between the various label names in the template document of the analysis report, template structure data can be obtained; a grouping template page is edited according to the content data of the analysis report and the template structure data; and all the grouping labels in the analysis report are eliminated, so the analysis report is obtained. In comparison with the prior art, the method for establishing the analysis report has the advantages that format setting and modification are more flexible, and maintenance is easier. In comparison with the prior art, the method provided by the invention has the advantages of the more flexible format setting and the easier maintenance.

The invention discloses a wind turbine generator maintenance work quality review auxiliary method and system. The method comprises the steps: taking all pictures in a work order as an analysis unit through the obtaining synchronization of picture data information and work order data, inputting each picture into a trained image recognition and classification model, and carrying out the recognitionand classification of all pictures in the work order; providing classification categories and confidence probabilities by a model, simultaneously performing md5 information mapping on each picture, storing picture names and mapping information into a database so as to repeatedly compare, finally judging whether the quality of the overhaul work order is qualified or not according to regular inspection specification requirements, and visualizing a result. According to the invention, the pictures collected in the daily maintenance process can be automatically classified, the work orders corresponding to the pictures can be automatically missed and analyzed, and the condition of repeatedly using the pictures can be automatically recognized. A report is automatically formed for the analysis result statistics, visual analysis is facilitated, and effective supervision is formed for key point pictures of work order missing detection (missing transmission) and the site where the pictures are repeatedly used.

The invention provides a method and device for processing DCP names and an image management system. The method comprises the steps that standard image information is acquired, copied information is set up through the standard image information, and the standard image information comprises a standard image name; the copied information matched with guide-in DCPs is determined; a binding relation is generated according to the DCPs and the matched copied information, and the standard image information bound with the DCPs is displayed according to the DCP. According to the method and device and the image management system, the aim of displaying the standard image information name, rather than a long cluster of characters, bound with the DCPs according to the DCPs is achieved, readability is improved obviously, the accuracy of user files is greatly improved, and the probability of error operations caused by the poor readability of file names is lowered.