128results about How to "Reduce the recognition rate" patented technology

The invention discloses an identity identification method based on self-established sample libraries and composite characters in video monitoring. The method comprises the steps that firstly, preprocessing is conducted on an acquired video, foreground detection is conducted, so that moving object information is obtained, then face detection is conducted on the basis of the moving object information, the detected face is identified, if the currently detected face can not be identified, a user is inquired so as to identify the detected face, and the identified face is added to the face sample library; if the face can not be detected in foreground information, pedestrian detection is conducted, a detected pedestrian is traced, gait period detection is conducted on a traced pedestrian image sequence, features of detected gait information of a period are extracted and identified, if identification fails, gaits are classified in the same mode of user identification and added to the gait sample library. The identity identification method provides a solution for identity identification on the condition of a lack of training sample diversity or small samples.

The invention provided a logistics code identification and sorting method based on multi-task deep learning. A method used for acquiring logistics code label images in all directions, a designing scheme capable of conveniently and visually detecting and positioning logistics code labels, a designing scheme suitable for character positioning and the segmented character size, character pattern and intervals between characters of the logistics code labels, a Faster R-CNN network used for detecting and positioning the logistics code labels, an algorithm module used for deviation rectifying of thelogistics code labels and detecting regular and reverse states of the characters, a character segmentation algorithm module used for conducting segmentation treatment on the characters on the logistics code labels, a multitask deep convolutional neural network used for deep learning and training identification, and a sorting control module used for controlling the sorting action according to the identified logistics code are included. According to the logistics code identification and sorting method based on multi-task deep learning, the problem that a large number of goods with irregular shapes and flexible packaging cannot be quickly and automatically sorted is effectively solved.

The invention provides a speaker recognition method combining a Gaussian mixture model and a quantum neural network. The method provided by the invention comprises the following steps: at the training stage, framing input training voice signals, extracting characteristic parameters and generating characteristic parameter vectors; then using a K mean value method and an EM (expectation-maximization) algorithm to obtain the Gaussian mixture model parameters of the characteristic parameter vectors, and finally utilizing the Gaussian mixture model parameters of all the training voice signals to train the quantum neural network; and at the recognition stage, obtaining the Gaussian mixture model parameters of a recognized speaker, then inputting the model parameters into the trained neural network, and obtaining a recognition result. The speaker recognition method is applicable to recognition of the speaker under the condition of less sample data and unbalanced sample data, and simultaneously the capability of the quantum neural network which can carries out effective judgment on voice data with cross data and fuzzy boundary of the speaker is utilized, so that the correct recognition rate of a system can be improved.

A logistics composite code identification method based on multitask depth learning comprises a method for acquiring logistics composite code images in all directions, a label design scheme convenientfor visually detecting and positioning a spray code character, a design scheme suitable for the character positioning of the label of the spray code character, a segmented character size, a font and an interval among the characters, a composite code design scheme suitable for visual identification, a Faster R-CNN network used for detecting and positioning a composite code, an algorithm module usedfor composite code image rectification and the forward / backward detection of the characters, a multi-task deep convolutional neural network used for deep learning and training identification, a convolutional neural network based on the character identification on the label of the spray code character of deep learning, the algorithm module used for identifying a one-dimensional barcode in the composite code, the algorithm module used for identifying the two-dimensional barcode in the composite code, and a sorting control module used for controlling sorting action according to identified composite code information. In the invention, a problem that a lot of randomly-placed, irregular-shaped, flexible packaging cargos can not be quickly and automatically sorted is effectively solved.

The invention discloses a single sample face recognition method in photo recognition. The single sample face recognition method comprises the following steps of: performing face reconstruction on all photo images in a training sample set, and obtaining a novel training sample set; performing Gabor character representation and data dimension reduction processing on the novel training sample set and images in the testing sample; and finally, performing face classification on the Gabor characteristics of the testing sample subjected to dimension reduction by using a support vector machine. According to the method, a complete and effective single sample face recognition solution is provided, the influences of illumination, posture and other factors are avoided to a certain degree, the method is high in recognition rate and high in robustness, and the influence of the age difference on the system performance can be lightened to a certain degree.

Concerned is speech recognition that reference speech information is extracted from a plurality of speech recognition dictionaries in a hierarchical structure to compare between extracted reference speech information and an inputted speech thereby recognizing the speech. Reference speech information representative of hierarchical-level skipping is prepared in a predetermined speech recognition dictionary so that, when recognizing an input corresponding to the reference speech information representative of hierarchical-level skipping, speech recognition is carried out by extracting a part of speech recognition dictionary belonging to a lower hierarchical level of the reference speech information being compared.

A pneumatic tire includes a surface and is rotatable about a center axis. The surface of the pneumatic tire includes a ground colored region including a surface of rubber, and an image recognition belt region including a colored region provided in a circumferential direction of the center axis.

The invention discloses a free-hand sketch offline identification and reshaping method. The method comprises the following steps of: preprocessing an input image; converting a discrete and disordered point set in each connected domain into an ordered point sequence for compression; fitting a plurality of straight lines of the point sequence by using a dynamic programming algorithm, and determining the number of optimum fitted straight lines to obtain the strokes, represented by the straight lines, of each connected domain; analyzing a stroke result obtained by fitting the plurality of straight lines; if the number of the fitted straight lines is greater than the maximum edge number of a shape identified by a system, performing order reduction on the strokes, classifying the strokes, and calculating distance between the strokes; and selecting the nearer strokes for combination and analysis, and performing verification according to the geometrical characteristic to determine the shape formed by combination of the input strokes. The method is higher in identification rate, and the shape to be identified has scale invariance and rotation invariance; and the algorithm supports a multi-stroke form for identifying limited shapes, and the problems in the identification completely based on the geometrical characteristic are solved.

The invention discloses a health index evaluating method and system for achieving animal behavior monitoring. The method comprises the steps that an RFID tag is put on each animal, and corresponding RFID tag collectors are arranged at animal activity points; the moving speed weighted value of each animal is obtained; the food scrambling capacity weighted value of each animal is obtained; the activity frequency weighted value of each animal is obtained; the sojourn time weighted value of each animal is obtained; the sum of the activity frequency weighted value and the sojourn time weighted value of each animal is obtained; the health index of each animal is judged based on the comprehensive weighted values of all the animals. According to the health index evaluating method and system, the animal behavior analyzing method adopted in the system is simple, practical and better suitable for collecting and analyzing mass data of large-scale breeding, the response of the system is accelerated, and the animal epidemic situation realizing time is shortened.

The present invention provides a road traffic sign detection and identification method. The method comprises the steps of: extracting a region of interest; performing multi-scale sliding traversal; merging image features; and identifying a traffic sign. The method specifically comprises: performing extraction of a region of interest of the traffic sign on an input to-be-detected image; convertingthe to-be-detected image into a gray-scale image; constructing a binary mask image of the region of interest of the traffic sign; performing multi-scale sliding traversal on the gray-scale image and the binary mask image to obtain position coordinates of a detected target; merging the extracted local texture features, local image region features and global features of the to-be-detected image; andclassifying and identifying the merged image features by using a classifier. The present invention relates to an electronic device and a readable storage medium for performing the road traffic sign detection and identification method; and the present invention also relates to a road traffic sign detection and identification system. The technical scheme of the present invention has a high detection rate, a high identification rate, a fast calculation speed, a low false detection rate and good robustness.

The invention discloses a radar range profile statistics and recognition method based on a PPCA model in the strong noise background, which relates to the technical field of radar automatic target recognition and mainly solves the problem that the current statistics and recognition methods based on the PPCA model are not robust to noises. The training phase comprises the following steps: framing, translating, aligning and strength-normalizing radar HPPR continuously, learning the parameters of each azimuth frame of the PPCA model by adopting the processed HRRP and storing a template. The test phase comprises the following steps: first strength-normalizing, translating and aligning the samples to be tested and then estimating the range of the signal-to-noise ratios (SNR) of the samples; computing the distance value of each frame of each target and deciding the category attribute if the SNR is more than 30dB, and rewriting the distance value, solving the noise energy under SNR condition by minimizing the distance value, finally computing the distance value of each frame of each target and deciding the category attribute if the SNR is less than 30dB. The method has the advantages of robustness to noises and less computation and is applied to identifying radar targets.

A method of compiling an image database for a three-dimensional object recognition including the steps of: when a plurality of images each showing an object from different viewpoint are inputted, extracting local features from each of the images, and expressing the local features using feature vectors; forming sets of the feature vectors, each set representing a same part of the object from a series of the viewpoints, and generating subspaces, each subspace representing a characteristic of each set; and storing each subspace to the image database with an identifier of the object to perform a recognition process that is realized by the steps of: when at least one image of an object is given as a query, extracting query feature vectors; determining the subspace most similar to each query feature vector; and executing a counting process to the identifiers to retrieve an object most similar to the query.

An individual-identifying communication system of the present invention comprises: plural light-emitting devices; an imaging device; a display device capable of displaying an image based on image data obtained by said imaging device; a selection device for selecting a predetermined area within the displayed image; a storage device storing plural identification information data for specifying a mobile terminal to communicate with, each of the identification information data being associated with a unique data indicating a lighting pattern of each light-emitting device; an acquisition device acquiring acquisition lighting pattern data from the image data from the imaging device; a determination device determining the unique data corresponding to the acquisition lighting pattern data, based on the acquisition lighting pattern data acquired from selection area image data and plural unique data; and a communication device communicating with a mobile terminal assigned with identification information data corresponding to the determined unique data.

The invention relates to an HMM-based part-of-speech tagging method and belongs to the field of information processing technology. According to the method, first, words in a word bank are ordered according to unicodes so that a dichotomy method can be used for quick search during word segmentation; second, an HMM is introduced, a tagged corpus serves as a training set and a test set to be used forobtaining three parameters of the HMM, and therefore a plurality of observable states in the HMM are obtained; third, secondary word segmentation is performed, the words not found in a primary word segmentation result are searched for in the observable states in the HMM, and a maximum entropy model is introduced to perform tagging on new words not found; and last, a viterbi algorithm is used to calculate an optimal hidden sequence of the HMM, and the optimal hidden sequence is combined with the tagging result of the maximum entropy model to obtain a final part-of-speech tagging result. Compared with the prior art, the phenomenon that a single part-of-speech tagging method is low in speed and low in new word recognition rate, and consequently a tagging result is low in accuracy is mainly solved, and the efficiency and accuracy of part-of-speech tagging are improved.

The invention discloses a radar range profile statistics and recognition method based on a FA model in the strong noise background, which relates to the technical field of radar automatic target recognition and mainly solves the problem that the current statistics and recognition methods based on the FA model are not robust to noises. The training phase comprises the following steps: framing, translating, aligning and strength-normalizing radar HPPR continuously, learning the parameters of each azimuth frame of the FA model by adopting the processed HRRP and storing a template. The test phase comprises the following steps: first strength-normalizing, translating and aligning the samples to be tested and then estimating the range of the signal-to-noise ratios (SNR) of the samples; computing the distance value of each frame of each target and deciding the category attribute if the SNR is more than 30dB, and rewriting the distance value, solving the noise energy under SNR condition by minimizing the distance value, finally computing the distance value of each frame of each target and deciding the category attribute if the SNR is less than 30dB. The method has the advantages of robustness to noises and less computation and is applied to identifying radar targets.

The invention discloses a non-contact human face identifying algorithm based on expanded eight-domain local texture features and an attendance system. The algorithm comprises the following steps of 1, extracting the expanded eight-domain local texture features of a human-face image, wherein the step comprises three stages, i.e. an image point marking stage, an image point encoding state and an image feature vector extracting stage; and 2, classifying the extracted local texture features by using an SVM (support vector machine) classifier, so that a human face is identified. The invention provides an expanded eight-domain local texture feature descriptor, which can describe the texture features in eight directions of a local region, and after identification rate of methods on an ORL (Optical Return Loss) database, an AR database and an FERET (Face Recognition Technology) database are compared, the obtained result shows that the descriptor provided by the invention is better than the other methods and particularly, the improved scheme has higher robustness.

The invention discloses an automatic identification method of foreign bodies in an electric energy meter based on acoustic detection, which comprises the following steps: collecting sound signal datain the electric energy meter; carrying out channel transformation on the collected sound signal data to extract the sound signal data containing foreign body channels; carrying out denoising processing on the extracted sound signal data; preprocessing the denoised sound signal data are preprocessed and forming an eigenmatrix, and obtaining the eigenvector corresponding to the maximum eigenvalue byprocessing the eigenmatrix. Feature vectors are input into the weak classifier of neural network based on Adaboost, and the feature vectors are used as the features of foreign body sound signals in the watt-hour meter for classification and recognition. The invention improves the detection efficiency of the electric energy meter, which is favorable for improving the automation process of the electric energy meter detection. The method shortens the detecting time of watt-hour meter, improves the production efficiency and the utilization rate of equipment, and realizes the fast, efficient, safeand reliable foreign body sound detecting work of watt-hour meter. The recognition rate of foreign object watt-hour meter is greatly improved.

Provided is a method for constructing an image database for object recognition, which includes a feature extraction step of extracting local descriptors from object images which are to be stored in an image database, a scalar quantization step of quantizing a numeric value indicating each dimension of each of the local descriptors into a predetermined number of bit digits, and a storing step of organizing each of the local descriptors after the quantization to be able to be searched for in the closest vicinity, giving to the local descriptor an identifier of the image from which the local descriptor has been extracted, and storing the local descriptor to which the identifiers are given in the image database. The storing step comprises extracting the local descriptors from the object images when a search query is given, scalar-quantizing each dimension, determining a local descriptor in the closest vicinity of each of the local descriptors from the image database, and storing each local descriptors so as to be able to identify one image by majority vote processing from the images including any determined local descriptor. The scalar quantization step comprises quantizing each dimension of each of the local descriptors into 8 bits or less. Also provided are a processing program for the method and a processing device for performing the processing.

The invention discloses an initiative gameguard network game system and a method. In the system, a player is required to choose two patterns with a same meaning from four characteristic patterns so that a natural person and a plug-in are distinguished. The system comprises a server portion and a client portion. The server portion is mainly used for generating a verification bottom pattern and characteristic codes of four verification patterns. The client portion converts the four characteristic codes into specific characteristic patterns, and the characteristic patterns are rendered on the bottom pattern and are displayed for the player. After the player clicks the two patterns with the same meaning by using a mouse, a client system sends a mouse clicking position to a server and the server determines whether it is the plug-in. The method is bound with a game content. A digital watermark technology and image generation technology are organically combined. Automatic identification of the plug-in can be effectively avoided and the plug-in can be high-efficiently identified.

The on-line foreign matter detecting process based on unit grey homogeneity includes the following steps: 1) image taking; 2) image transmission: 3) foreign matter distinguishing and processing; 4) distinguishing signal transmission; and 5) eliminating foreign matter. During the process, LED is used as light source to raise image brightness and stability, images are taken from both the upper side and the lower side of the detected matter, and the detected matter is distinguished from two sides for raised distinction rate. In the foreign matter distinguishing and processing, the integral unit gray homogeneity is utilized in judging the unit attribute to distinguish the foreign matter with color similar to that of the detected matter. The present invention has raised distinguishing rate between tobacco or other detected matter and foreign matter.

The embodiment of the invention provides a face image recognition method and device, belonging to the technical field of image recognition and aiming to improve the face image recognition rate. The face image recognition method comprises the following steps: matching the characteristics of the first face image acquired through the near-infrared light and the face image in the preset near infrared light registration set and classifying to obtain the categories and the matching scores of the top M candidates with the highest of matching scores in the first group; matching the characteristics of the first face image acquired through the visible light and the face image in the preset visible light registration set and classifying to obtain the categories and the matching scores of the top M candidates with the highest of matching scores in the second group; normalizing the matching scores of the top M candidates with the highest of matching scores in the first group and the matching scores of the top M candidates with the highest of matching scores in the second group respectively; and merging the same categories to obtain the N categories and the scores which corresponds to the N categories, and taking the categories with the highest scores as the recognition results. The face image recognition method and device are mainly used for face image recognition.

A speech recognition device presenting whether a user's utterance is an unregistered word and whether the utterance should be repeated. The device includes a vocabulary storage unit (102) defining a vocabulary for speech recognition, and a speech recognition unit (101) checking the uttered speech against registered words. The device also includes a similarity calculation unit (103) calculating a similarity between the uttered speech and acoustic units, a judgment unit (104) judging, based on the check by the speech recognition unit (101) and the calculation performed by the similarity calculation unit (103), whether the uttered speech is a registered or unregistered word, an unregistered word unit (106) storing unregistered words, an unregistered word candidate search unit (105) searching the unregistered word unit (106) for unregistered word candidates, the, when the judgment unit (104) judges that the uttered speech is an unregistered word, and a display unit (107) displaying the result.

The invention discloses a novel digital extraction and identification method and a system thereof. The method comprises the following steps that a mobile electronic device identifies and processes handwritten digits through scanning; through an OSTU algorithm, a scanning result is changed into black and white; through a vertical projection segmentation algorithm, a series of digits are segmented into independent individuals; through an Hilditch algorithm or an ZS algorithm, the digits are refined; through bilinearity or a bicubic interpolation algorithm, a refined digital image is zoomed; through a rough grid characteristic extraction method, structure characteristic extraction is performed on the normalized digital image so as to generate a characteristic vector and input into a recognizer; a trained SVM recognizer is used to classify and identify the input characteristic vector so as to acquire an identification result of the digital image; and combining the single identification result to form a character string and output to a user interface. Through scanning, the handwritten digits are recorded into the electronic device, which is convenient for a next operation. The method and the system are convenient and rapid.

A traffic sign recognition method analyzes and classifies image data of a sensor in an information processing unit. The image data is analyzed to select an image portion judged to contain a traffic sign of a particular sign class. A class-specific feature is identified in the image portion. A modified image portion is created, in which the class-specific feature has been shifted to a center of the modified image portion. Then the modified image portion is evaluated by a learning-based algorithm to recognize the traffic sign of the particular sign class.

A method of controlling a photographing apparatus includes: recognizing a gesture in an input image as a previously defined gesture; if the gesture corresponds to a previously defined gesture for a zoom operation, performing the zoom operation using optical zoom; and while the zoom operation is being performed, continuing recognizing the gesture by taking into consideration the zoom operation.

Magnetic toner particles are disclosed. The magnetic toner particles contain at least one polymeric binder and at least one magnetic additive, wherein the surface of the toner particle contains particles of positively chargeable inorganic fine powder particles. The inorganic fine powder particles have a mean volume average particle size of from about 0.5 to about 7 mum, and a cleaning ratio of from about 0.1 to about 5.0 and a cleaning ratio being the volume fraction of particles between 0 and 1.0 mum, divided by the volume fraction of particles greater than 1.0 mum; and the toner particles having on the surface thereof a flowability improving agent having a BET surface area of at least about 30 m<2> / g. Methods of forming electrostatic images are further disclosed. Also, images formed from the magnetic toner particles are further disclosed and have excellent character void frequency, total void area, and suitable magnetic signal strengths. Developers containing the magnetic toner particles of the present invention are also disclosed.

The invention discloses a motor imagery electrocorticogram (EEG) signal classification method based on independent component analysis, which comprises the following steps: S1, collecting the EEG signals and preprocessing the EEG signals, and randomly dividing the preprocessed EEG signals into a training set and a test set; S2, sequentially selecting the training set data to carry out independent component analysis and calculation on the single test sample data, and automatically identifying and acquiring the motion related component based on the spatial distribution pattern of the source; S3:classification and recognition of motor imagery based on zero training classifier; S4: optimizing the selection of leads by using the training set data, substituting the optimized leads into the testset, and looping the steps S2 and S3 to obtain the final classification recognition rate. The invention can reduce the spatial model matching problem caused by the difference between the collected EEGdata, and has high recognition accuracy to the motor imagery EEG signals.

Provided is an input and output integrated module for simultaneously linking biometric information algorithms, which includes an integrated module for simultaneously authenticating fingerprints and finger veins which includes a fingerprint module, a finger vein module, and a conversion module, wherein: the fingerprint module scans a fingerprint in an image sensor, compares the scanned image with a previously-stored image, and outputs a registered authentication code of a corresponding person as a specific serial communication signal when there is a person having a fingerprint identical to the previously-stored image; the finger vein module scans a finger vein in a camera image sensor after light of an infrared light-emitting diode passes through a finger, compares the scanned image with a previously-stored image, and outputs a registered authentication code of a corresponding person as a specific serial communication signal when there is a person having a finger vein identical to the previously-stored image.

The invention discloses a resource downloading system and method, and a crawler downloading system. The resource downloading system limits a capture frequency of a crawler to a site through a token bucket flow limiting method; a value range of a timestamp increase value of a token generated by a random token generator is determined according to the rate of query per second of the crawler to the site; and the timestamp of the token is determined according to a timestamp at a current moment and the timestamp increase value, so that the timestamp increase value of the token is limited to be a random value, the capture frequency of the crawler to the site is limited to be a random frequency, an anti-crawler program is prevented from identifying the crawler according to the characteristic, namely, a fixed capture frequency, the probability of identifying the crawler by the anti-crawler program is reduced, and the functional stability of the crawler is improved.

Various embodiments of the present document relate to a display and, particularly, relate to a display including infrared elements, and an electronic device including the same. The display according to various embodiments of the present document includes a substrate; a first element for outputting light of a first visible light band and a second element for outputting light of an infrared band, the elements being formed on the substrate; and a third element for outputting light of a second visible light band, wherein at least a portion of the third element is arranged to overlaps the second element.