85 results about "Image pattern recognition" patented technology

Systems and methods for image pattern recognition comprise digital image capture and encoding using vector quantization (“VQ”) of the image. A vocabulary of vectors is built by segmenting images into kernels and creating vectors corresponding to each kernel. Images are encoded by creating a vector index file having indices that point to the vectors stored in the vocabulary. The vector index file can be used to reconstruct an image by looking up vectors stored in the vocabulary. Pattern recognition of candidate regions of images can be accomplished by correlating image vectors to a pre-trained vocabulary of vector sets comprising vectors that correlate with particular image characteristics. In virtual microscopy, the systems and methods are suitable for rare-event finding, such as detection of micrometastasis clusters, tissue identification, such as locating regions of analysis for immunohistochemical assays, and rapid screening of tissue samples, such as histology sections arranged as tissue microarrays (TMAs).

The present invention relates to a method for correcting at least one color of a photographic image including at least one pattern area or image pattern with a predictably known color or memory color, said image being transferred to a digital representation, wherein the method comprises the following steps: said at least one pattern area or image pattern is being detected with respect to its presence and its location, and preferably also with respect to its dimensions; an existing color in the at least one detected pattern area or image pattern is being determined; at least one replacement color value (memory color) is being provided, said value being related to the respective at least one pattern area or image pattern and the determined existing color is replaced by said at least one replacement color value, to correct the color in the image pattern or image area.

The invention relates to a robust ear detection method which improves an AdaBoost arithmetic and belongs to the technical field of image mode identifying. The invention is characterized by proposing an ear detection method with excellent performances under a complex underground. The invention proposes four anisomerous Haar-like corner characteristics which are used for describing the grayscale changes on the partial areas of the ears; a policy of subsection selection is adopted for selecting the best sorting threshold of the Haar-like characteristics, thus reducing the sample training time; the weight of a weak sorter is modified for reducing the mistaken detection rate of the sorter; the threshold HW is set and eliminated according to the distribution change of the sample weight in the training, thereby preventing an over-studying phenomenon from being generated and leading the miss-detection rate and the mistaken direction rate of the ear detection to be reduced; besides, the invention also provides a single-ear detection policy for leading both the defection efficiency and the detection effect to be improved. The excellent performances of the robust ear detection method are shown on a PC machine and a DSP.

The invention discloses a refrigerator. The refrigerator comprises a camera and a control device, wherein the camera is arranged on the refrigerator and is used for acquiring a motion process image showing that a user puts food in or takes the food out; the control device is used for performing image segmentation and image mode identification on the motion process image to determine a motion type and corresponding food information, wherein the food information comprises the name and storage position of the food; managing statistical data of the food in the refrigerator according to the motion type and the corresponding food information. The name, storage position and the like of the food which is put in or taken out by the user can be intelligently and automatically identified in the door opening period of the refrigerator by using only one camera, and the refrigerator is higher in timeliness, better in user experience and low in cost. The invention further discloses a method for managing the food in the refrigerator.

The invention discloses a fire image depth learning pattern recognition method, comprising the following steps: preprocessing a test sample and a training sample of the fire image; designing an improved GoogleNet-based deep learning network, which extracts features automatically through training to eliminate artificial traces; deep features are used to train the classifier and identify the fire test samples. The invention applies the image multi-scale convolution fusion network to the fire image pattern recognition technology, remarkably improves the training efficiency of the network, effectively solves the problems encountered in the fire recognition method, such as the accuracy and real-time property are not ideal, the network structure is complex, and the training time is long, the stability and the robustness are poor, and the like. The recognition accuracy of the trained network is 99.2% in three types of fire images.

Provided is a monitoring method for electric power working safety based on image recognition. A structure of the method of the invention comprises an on-site electricity-utilization controllable operation power supply, a camera and a system respectively connected with the on-site electricity-utilization controllable operation power supply, wherein the system is composed of an image pattern recognition processing program and has video monitoring and electricity-consumption environment control functions. The monitoring method comprises the steps of realizing a virtual boundary for an electric power working operation site and realizing three-level safety control and environmental electricity-utilization control based on an image recognition technology so as to prevent workers at the electricpower working operation site from electric shocking and to monitor operation conditions of on-site electric power equipments. By conducting effective safety boundary management of an electric power working and testing site, the method of the invention has the substantial advantages of improved functions, reliability and wide application range.

The invention relates to a cross-modal image mode identification method for the biomedical literature, belongs to the field of image identification and natural language processing and aims to solve across-modal image identification problem in the medical literature. The method is characterized in that an image and a text are processed through utilizing a pre-trained cross-modal composite image detection model, and whether the to-be-identified image is a composite image is determined; the image and the text are processed through utilizing a pre-trained cross-modal composite image multi-label classification model to output a biomedical pattern category of a composite image sub picture; a pre-trained cross-modal simple image pattern classification model is utilized to process the image and the text to output a biomedical pattern category of simple images. The method is advantaged in that image pattern identification tasks in the biomedical literature are effectively completed, resourcesin the general and biomedical fields are utilized to improve the identification performance, and labor and time cost is reduced.

The invention relates to an optical fiber perimeter intrusion monitoring method based on image recognition. The monitoring method is completed by an optical fiber vibration sensing system. The system comprises detection cables which are connected together, a monitoring host and an upper computer. The monitoring method comprises the steps that vibration caused by outside intrusion is detected by the detection cables, and an optical signal is transmitted to the monitoring host; the monitoring host converts the received optical signal into an electrical signal, and then carries out sampling and analog-digital conversion on the electrical signal to acquire a discrete digital signal; the digital signal is transmitted to the upper computer; the upper computer processes the acquired digital signal to acquire processed signal feature amount to form one or more waterfall curves; according to the form of the waterfall curve, image pattern recognition is carried out; and an intrusion event is determined, intrusion alarm is triggered. According to the optical fiber perimeter intrusion monitoring method provided by the invention, additional equipment is not needed; real-time intrusion recognition is realized without a storm database; and the method has the advantages of high intrusion recognition rate and low false alarm rate.

A system that incorporates teachings of the subject disclosure may include, for example, a processor that can detect an event, access location information for a group of mobile communication devices that are each automatically capturing images, and identify a subset of the group of mobile communication devices that are in proximity to the event based on the location information. The processor can provide first image analysis criteria to the subset of the group of mobile communication devices without providing the first image analysis criteria to remaining devices of the group of mobile communication devices where the first image analysis criteria includes first characteristics associated with an object. The processor can receive a first target image that includes the object from a first mobile communication device of the subset of the group of mobile communication devices, where the first target image is selected by the first mobile communication device from among a plurality of images captured by the first mobile communication device based on first image pattern recognition performed by the first mobile communication device utilizing the first image analysis criteria. Other embodiments are disclosed.

The invention relates to a citrus pest monitoring early warning method based on a distributed compression perception WSN (wireless sensor network). The distributed compression perception and wireless sensor network is applied to environment monitoring on citrus crops and early warning of citrus pests. The method comprises steps that combined sparse expression, code measurement and combined information reconstruction of the image data monitored by wireless sensor network nodes are carried out through utilizing the distributed compression perception technology; a deep convolution neural network (DCNN) algorithm is utilized to carry out image mode identification, and a pest parameter (image) characteristic database is established; the citrus pest bioinformation and environment information parameter indexes are quantified, and pest disaster early warning is timely and accurately emitted according to a decision mechanism. Through the method, the distributed compression perception image processing technology is utilized, transmission pressure of large-volume images in the wireless network is reduced, a timely, on-site, low-power and strong-vitality wireless network monitoring system is established through utilizing the wireless sensor network, agricultural monitoring precision is improved, and a citrus pest early warning period is shortened.

The invention discloses a lip-shaped identifying sounder, and its connecting mode: a micro camera head is connected with an image collecting unit, the output end of the image collecting unit is connected with a lip-shaped image mode identifying unit, the signal of the identifying unit is outputted to a voice synthesizing unit, the synthesizing unit is connected with a voice storage unit, the synthesizing unit extracts voice synthesizing elements from the voice storage unit to synthesize a sound signal and output the signal to a sounding unit. Then a speaker gives out a corresponding sound to the lip shape and its variation sequence. By identifying the lip shape of a speaker, it determines speech contents, makes voice synthesis on the speech contents and real-timely gives out a sound by a speaker. It can help the persons unable to sound because of removal of throat or vocal cords or the deaf-mute able to speak lip languages to sound, convenient for them to exchange with the normal persons.

The invention discloses a red light running detecting device and method based on video processing and belongs to the fields of image pattern recognition, video target tracking, intelligent video monitoring and intelligent transportation. The red light running detecting method comprises a license plate locating method based on edge information and color information and a vehicle target tracking method based on feature points of an angular point. According to the red light running detecting method, the most obvious partial characteristics of an image of a vehicle target are used as the basis of vehicle locating, relevance of the characteristics of a rigid body of the moving vehicle of multiple frames is used for tracking the vehicle, and therefore movement information of the vehicle target in a monitoring video can be accurately obtained. The red light running detecting device and method based on video processing can rapidly and accurately detect the position of a license plate of the vehicle in a monitoring video, can accurately obtain the movement information of the vehicle target, can intelligently achieve the function of vehicle running snapshot, can intelligently achieve the function that whether the vehicle runs on a required lane is judged, and is applicable to a red light running electronic police system, a public security access system, an abnormal-vehicle event detection and no-parking automatic fare collection system and the like.

A computerized directed feature development method receives an initial feature list, a learning image and object masks. Interactive feature enhancement is performed by human to generate feature recipe. The Interactive feature enhancement includes a visual profiling selection method and a contrast boosting method.A visual profiling selection method for computerized directed feature development receives initial feature list, initial features, learning image and object masks. Information measurement is performed to generate information scores. Ranking of the initial feature list is performed to generate a ranked feature list. Human selection is performed through a user interface to generate a profiling feature. A contrast boosting feature optimization method performs extreme example specification by human to generate updated montage. Extreme directed feature ranking is performed to generate extreme ranked features. Contrast boosting feature generation is performed to generate new features and new feature generation rules.

The invention relates to an automatic classification technology of solid particles based on an image pattern recognition technology, belonging to the field of image processing technology, in particular to an embedded bean classifying system which can be used for progressively analyzing the composition of beans and classifying mixed beans with similar colors and is suitable for various common beans. The whole classifying system can be integrated in a limited small region through the embedded design to meet the strict restriction of bean processing devices such as bean milk machines on volume, thereby possibly realizing highly automatic machining processes of various beans in a same device.

The invention discloses a device and a method for detecting the position of an object. The device comprises a bracket, an image acquiring device and a video projecting device, wherein the bracket, the image acquiring device and the video projecting device are arranged above an object to be detected; the video projecting device is arranged on the bracket and is adjustable in projecting angle; and the image acquiring device is used for acquiring a projected image on the surface of the object to be detected and is arranged on the bracket and connected with a computer through a communication cable. The method comprises the following steps: using the video projecting device to project a concentric circle or other figures on the object to be detected; using the acquired projecting image to adjust the angle of lenses of the projecting device and the image acquiring device and making the angle perpendicular to the object to be detected; because the size of the image projecting on the object changes along with the change of the position of the object to be detected, acquiring an image changing along the change of the position by the image acquiring device; and finally processing the image by a computer picture mode recognition system to calculate the position of the object. The device and the method have the characteristics of low cost, no need of defending, strong anti-interference performance, convenient installation and debugging, and simple and convenient operation.

The invention discloses a transformer integrated temperature monitoring device for infrared image recognition, and relates to the technical field of comprehensive monitoring for grid power transformers. The transformer integrated temperature monitoring device is characterized in that an embedded system control core unit is connected with a touch screen; and an image acquisition module is connected with the embedded system control core unit. The control core unit of the transformer integrated temperature monitoring device acquires an infrared image of a setting range of a transformer top view region from the image acquisition module, retrieves the image according to a set infrared temperature threshold and a pixel change threshold and transmits the image to a cloud platform through the core control unit and an embedded GPRS / DTU module when exceeding the thresholds. A micro-service function of the cloud platform is for image pattern recognition. If the temperature exceeds the threshold, threshold-exceeding equipment and the fault type are recognized, early-earning information and handling suggestions are transmitted to the operation and maintenance personnel; and if the personnel get close to the transformer, a camera collects a visible light image and uploads the image to the cloud platform for behavioral identification, LED lamplight warning and voice warning are performed if there is a dangerous behavior, and warning information is sent to the operation and maintenance personnel at the same time.

The invention provides a bootstrapping weak learning method based on a random fern and a classifier thereof, which belong to the technical field of computer graphic identification. In graphic identification, positive and negative patterns are generally judged by adopting the mean value distances of weighting and Gaussian probability distribution of a weak classifier. Or classification trees are taken as weak learners, nodes are partitioned by using the partitioning rule of error metric reduction maximization, and the weak classifiers are improved into strong classifiers. However, the weak learning methods have the defects of low convergence rate, poor accuracy and low computing efficiency. Through steps for selecting image characteristics, constructing a random fern and a weak learning method based on a random fern, establishing a weak classifier and a result classifier and the like, the image mode identification problems of complexity of an imaging environment and strict requirement on the operand can be well solved, a rapidly-converging and efficient bootstrapping weak learning method is realized, and a high-accuracy classifier for real-time processing is obtained. The method and the classifier are mainly applied to identification occasions in various modes.

The invention discloses a knife switch state closing reliability judging method based on the distance between knife switch arm feature points, which achieves accurate positioning of a contact area of a knife switch in an image in a feature point matching mode, achieves recognition positioning of circle feature areas respectively located on two knife switch arms in an image circle detection mode, calculating the diameter of a circle on any side and the distance of two circle area centers to calculate the proportionate relationship between the diameter and the distance so as to achieve knife switch state closing reliability judging. The knife switch state closing reliability judging method based on the distance between knife switch arm feature points has the advantages of achieving knife switch closing reliability judging based on the image pattern recognition technology, replacing operators to achieve knife switch closing in place reliability judging, greatly reducing labor intensity of the operators, shortening operation time, eliminating influence of subjective factors of onsite staff, and being small in working load and system cost and high in practicality.

The invention discloses a method based on the contact proportion of a dynamic contact and a static contact and used for distinguishing switching-on and switching-off reliability of a disconnecting link. The method comprises the following steps: step 1. establishing a standard data template for the proper switching-on operation according to data information of the concrete position of the contact of the disconnecting link when the disconnecting link is properly switched on; step 2. analyzing by a computer the collected visible light image information or infrared image information about the switching-on condition of the disconnecting link collected by a mobile robot at real time; and step 3. comparing the switching-on condition of the disconnecting link collected at real time with the established standard data template for the proper switching-on operation, if the switching-on condition of the disconnecting link is consistent with the standard data template, then the disconnecting link is reliably switched on into place; otherwise, the disconnecting link is not reliably switched on into place. The method provided by the invention has the beneficial effects of realizing identification on the switching-on and switching-off reliability of the disconnecting link on the basis of an image pattern recognition technology, operators can be replaced to realize identification on the switching-on and switching-off reliability of the disconnecting link, the labor intensity of operators is greatly reduced, the operation time is shortened, and the influences of the subjective factors of on-site personnel are removed.

Disclosed are a quantum system-based image pattern recognition computation apparatus and method for machine vision and a quantum system-based machine vision apparatus. The computation apparatus recognizes patterns between images in machine vision by using a quantum system. The computation apparatus includes a modeling unit and an interpretation unit. The modeling unit sets up an objective function based on the similarity between a first pattern derived from the relationships between points of interests of a first image and a second pattern derived from the relationships between points of interests of a second image. The interpretation unit finds an optimum first pattern and an optimum second pattern, in which the similarity between the first pattern and the second pattern is optimized, by interpreting a final quantum state obtained through an adiabatic evolution process of the quantum system in which the objective function is optimized.

The invention relates to a stage light following control system, and particularly to a stage intelligent light following control system and a stage intelligent light following control method based on image pattern recognition. The stage intelligent light following control system and the stage intelligent light following control method settle the problems of difficult arrangement, complicated wiring, application range limitation, high application cost, low controlling real-time performance, and low accuracy. The stage intelligent light following control system based on image pattern recognition comprises the components of a high-resolution camera, an upper PC, stereo audio equipment, a first RS485 communication machine, a second RS485 communication machine, a third RS485 communication machine, a microprocessor, a pan / tilt spotlight system and an adjustable light system, wherein the output end of the high-resolution camera is connected with the input end of the upper PC through an audio line. The output end of the upper PC is connected with the input end of the stereo audio set through an audio line. The upper PC is bidirectionally connected with the first RS485 communication machine. The stage light following control system is suitable for stage light following control.

A system that incorporates teachings of the subject disclosure may include, for example, a processor that can detect an event, access location information for a group of mobile communication devices that are each automatically capturing images, and identify a subset of the group of mobile communication devices that are in proximity to the event based on the location information. The processor can provide first image analysis criteria to the subset of the group of mobile communication devices without providing the first image analysis criteria to remaining devices of the group of mobile communication devices where the first image analysis criteria includes first characteristics associated with an object. The processor can receive a first target image that includes the object from a first mobile communication device of the subset of the group of mobile communication devices, where the first target image is selected by the first mobile communication device from among a plurality of images captured by the first mobile communication device based on first image pattern recognition performed by the first mobile communication device utilizing the first image analysis criteria. Other embodiments are disclosed.

The invention relates to a multi-group image supervised classification method based on empirical mode decomposition, belonging to the image processing field. The multi-group image supervised classification method solves the problems that the feature utilization is not enough and the classification accuracy is low in the existing supervised classification method. The method comprises the following steps: 1, carry out empirical mode decomposition to a feature vector of each pixel and obtain an expanded feature vector; 2, combine the original feature vector and the expanded feature vector thereof of each pixel according to an uniform rule and obtain an expended dimension feature vector; 3, carry out training to a support vector machine, then carry out belonging judgment to the categories of an expended dimension testing feature vector, and form a plurality of support vector machine sub classifier; 4, build a multi- classifier based on an one-to-one policy to make decisions to the belonging categories of each pixel, and finish the category of the multi-group image. The multi-group image supervised classification method is used for multi-group image pattern recognition which needs high-precision category.

The invention relates to a method for using a laser radar scanning method to calibrate dynamic pedestrians and vehicles in a video in a snowing or raining state, relates to the field of image pattern recognition and machine intelligence, and solves the problem that an existing algorism is poor in real-time performance and inaccurate in recognition of dynamic objects. When a raining or snowing state is detected in a video, the video is decomposed into frames of images; a laser radar is utilized to scan the front of a video window, distance information of dynamic objects is measured through the laser radar, instantaneous speed of the dynamic objects is obtained, the types of the dynamic objects are judged according to the instantaneous speed, and the types of the dynamic objects include pedestrians and vehicles; center line positions of the objects are scanned through the laser radar, and position coordinates of the pedestrians and vehicles in the video are obtained according to the center line positions; the position coordinates of the pedestrians and vehicles in the Step 3 are mapped to the images in the Step 1, thereby obtaining mapping images; and an edge-preserving denoising method is adopted to perform distortion-preventing processing on the mapping images, thereby obtaining distortion-preventing mapping images. The method provided by the invention is suitable for detecting pedestrians and vehicles.

According to the invention, a CCD paper defect detection system based on image pattern recognition is designed according to a paper quality grading method. According to the invention, with a CCD technology, paper surface quality is subjected to real-time detection and dynamic measuring; the paper products are automatically classified according to the qualities, and paper defects such as black spots and holes on the paper surfaces are detected; the papers are automatically classified, and alarming is performed when unqualified papers are detected; and a failure arte of the papers and a proportion of each grade of the papers can be obtained by automatic calculation. Therefore, detection efficiency is improved, and various data of the papers is automatically counted.

The invention discloses an esophageal cancer lesion area search and identification modeling method based on evolutionary neural network structure, and relates to the technical field of image mode recognition and medical images, and the method comprises the following steps: S1, collecting and labeling an esophageal image data set for training a neural network model; S2, constructing a neural network structure search space for esophageal cancer lesion area identification; S3, training a super-network model for esophageal cancer lesion area identification; S4, searching an optimal neural networkstructure on the constructed super-network model by using an evolutionary algorithm; and S5, finely adjusting the searched neural network structure, and predicting a lesion area on the newly input esophageal image. The dependence of neural network structure design on expert experience in an esophageal cancer intelligent recognition task is eliminated, so that the deep neural network method is easier to use in the aspect of esophageal cancer lesion area recognition.

The invention discloses a method and system used for recognizing and controlling a gesture and based on a hand profile feature. The method is mainly formed by an initial setting step, an image pre-processing step, a gesture recognizing step, a command parsing step and a controlling step for an application program to be controlled. Gesture recognition is conducted by extracting an outer profile of a video frame, the gesture is recognized according to gesture feature area and gesture feature perimeter in a gesture recognition process, a fist and a palm are described by using the sum of convex hull flaw area, and accuracy and efficiency of palm recognition are effectively improved. In the gesture recognition process, only common computational discrimination such as perimeter and area are used instead of a complex image mode recognition sample comparison method, so that an extra sample base is not required to be added and can develop service for other applications such as a picture viewer, a music player, a webpage browner and a game application program which are based on gesture control.

The invention belongs to the technical field of navigation and positioning based on image pattern recognition and computer, and discloses a rendezvous and docking visual navigation and positioning method based on coded targets. The present invention obtains the three-dimensional coordinates of the template points and feature points on the coded target under the CCD1 coordinate system through the design and identification of the coded target, and calculates The conversion parameters between the target spacecraft coordinate system and the CCD1 camera coordinate system are obtained, combined with the installation parameters of the CCD1 camera on the tracker, the relative position and attitude of the target relative to the tracker are obtained. Adopting the present invention to carry out navigation and positioning in the rendezvous and docking approach section does not require the radar navigation results of the short-distance guidance section, and the recognition results are not affected by the radar navigation results; the present invention can be used to produce different numbers of coded targets and install them on the space station. The accuracy and reliability of relative positioning between the tracker and the target can be improved.

The invention discloses a vehicle theft prevention device, belonging to the vehicle security monitoring technical field, which comprises a controller, an on-vehicle part and a safety guard part, wherein the controller is kept by vehicle owner for emitting locking or unlocking command to the on-vehicle part, the on-vehicle part is mounted on a vehicle for receiving the operation command of user and deciding to emit vehicle identity information to the safety guard part or not according to the conditions, and the safety guard part is mounted at the outlet of district or plane area. When the vehicle is thieved, thief can not operate the on-vehicle part to emit correct vehicle identity information, and when the thieved vehicle reaches outlet, the safety guard part recognizes the vehicle via image mode recognition technique, since received vehicle identity information or vehicle identity is incorrect, the safety guard part determines that the vehicle owner is suspected or even illegal.

The invention belongs to the Internet oral health field and discloses a tooth virtual editing method and system. The method includes the following steps that: the three-dimensional image basic database of human oral cavities is constructed; the three-dimensional image data of the oral cavity of a user are acquired, image pattern recognition is performed on the three-dimensional image data on the basis of the saved three-dimensional image basic database, and an object to which the image data belong is identified, and an object labeling system for the oral cavity of the user is established, wherein the object contains a block and / or an entity; and the object is virtually edited based on the established object labeling system for the current user, and the three-dimensional image of updated teeth is displayed. With the method and system of the invention adopted, he problem that a user cannot intuitively edit teeth in the prior art can be solved.