12 results about "Identification rate" patented technology

The invention discloses a coal-rock identification method based on image gray level co-occurrence matrixes. The coal-rock identification method comprises the following steps of: respectively collecting color images f1 and f2 of a known coal sample and a known rock sample under the same imaging condition; respectively capturing sub-images s1 and s2 which are the same in size and free of a background; extracting gray level co-occurrence matrixes P1 and P2 of the sub-images; calculating image characteristics K1 and K2 based on the gray level co-occurrence matrixes; for an unknown coal-rock object to be identified, collecting an image fx under the same imaging condition; capturing a sub-image sx free of the background according to the same size; extracting a gray level co-occurrence matrix PXof the sx; calculating an image characteristic value Kx based on the gray level co-occurrence matrixes; and judging the coal-rock type according to a relation between Kx and K1 and K2. Because non-contact identification based on the images is utilized, the coal-rock identification method has the characteristics of easy deployment, strong adaptability, high identification rate and the like.

The invention discloses an SAR image classification method for an SVM classifier by using a hybrid kernel function based on wavelet properties, belonging to the technical field of image processing and mainly solving the problem that effectiveness in image characteristic extraction is not sufficient. The method comprises the following steps: firstly, imputing training and testing sample images, and normalizing and marking the sample images; secondly, decomposing the normalized sample images, respectively extracting a plurality of characteristics from various decomposed sub-zones, and storing the characteristics in terms of a structural body of T1 x r; thirdly, according to the characteristics of the sub-zones, constructing the hybrid kernel function based on wavelet properties for the SVM classifier (see the formula on the lower right side, wherein, in the formula, Xi and Xj respectively indicate an ith sample image and a jth sample image, i and j are both less than or equal to 1, xik and xjk respectively indicate the kth chacteristics of the ith sample image and the jth sample, and Rhok is a convex combination coefficient; and fourthly, finishing the classification of the image characteristics by optimizing the convex combination coefficient in the hybrid kernel function. The method has the advantage of high image classification recognition rate and can be used for machine learning and mode identification.

The invention provides a finger vein identifying method based on characteristic value normalization and bidirectional weighting. The method comprises the following steps of: (1) collecting finger vein images by an image collecting device; (2) carrying out preprocessing on the collected finger vein images, wherein the preprocessing comprises the steps of graying the color images, extracting finger areas, respectively eliminating salt and pepper noise and Gauss noise by adopting a combined filter, segmenting and binarizing the images by adopting a local dynamic threshold algorithm, de-noising by adopting an area eliminating method, extracting the finger vein venation images by finger contour markers, and standardizing the sizes of the images into that of the uniform images; (3) extracting a finger vein characteristic by a two-dimensional main component analyzing algorithm with weighting in two directions of the line; and (4) matching and identifying by a most adjacent classifier. The invention is used for a finger vein identifying system, obviously enhances the identifying speed of the finger veins, and has stable and high identifying rate.

The scheme relates to artificial intelligence, and provides an emotion recognition method, a device and a computer readable storage medium, andthe method comprises the following steps of: receiving audio data, generating a voice spectrum graph aiming at the audio data, inputting the voice spectrum graph to a first recognition module, inputting the audio data to a second recognition module, whereinthe first recognition module extracts a first feature vector by adopting a DCNN network and an RNN network which are sequentially connected the second recognition module extracts a MFCC from audio data, andthe MFCC is transformed into a second feature vector by s non-linear transformation;and the first and second feature vectors output from the first and second recognition modules are connected to form a combined feature and be input into a full connection layer and a soft Max layer in turn for emotional recognition.. By means of the scheme, the combination of the DCNN and the RNN is beneficial for simulating subtle local emotion clues, the MFCC describes voice characteristics in a short time, more emotion related features are captured through two features extraction modes, and the emotion recognition rate is improved.

The invention provides a target detection identification method and system in a low altitude and sea surface environment. The method comprises steps of establishing a space virtual reference coordinate system and carrying out self-calibration; for different measurement conditions, using corresponding coordinate conversation methods, allowing data of measurement points and targets to be establishedin the space virtual reference coordinate system; calculating all geometrical parameters of each target and coordinates of the target; through two radars in different heights, carrying out self-calibration and blind supplementation; by use of a de-noising method of interframe correlation, processing radar signals; and carrying out information fusion processing on multi-source target data in the radars and carrying out target detection and tracking. According to the invention, by use of dynamic self-calibration technology, and adaptive detection and recognition technology, strong sea clustersare inhibited; positioning precision and recognition rate of the targets are improved; target detection ability for observing sea area during day and night in a severe weather environment is improved;and effective detection, tracking, recognition and law enforcement monitoring evidence obtaining for small targets on the sea surface and low-altitude small targets like unmanned aerial vehicles areachieved.

The invention provides a light source control system based on a double-light-source gun-type camera. The double-light-source gun-type camera transmits a video shot in real time to an AI module; the AImodule identifies the received video according to a preset identification material library to obtain a video identification result; the photosensitive sensor transmits the output ADC value to the lamp control module; and when the lamp control module recognizes that the current light becomes weak or enters the night according to the ADC value, the working state of the light source in the double-light-source gun-type camera is adjusted according to the video recognition result and the exposure of the double-light-source gun-type camera. The light source control system intelligently and continuously adjusts supplementary lighting according to videos of the double-light-source gun-type camera, uses a specific supplementary lighting effect according to whether specific personnel, behaviors orpostures are recognized or not, and continuously and dynamically adjusts according to video data, so that the user experience is improved, and the false alarm rate is reduced. The AI module of the system is automatically updated, so that the recognition rate is improved, and the safety coefficient and the user experience are improved.

The invention provides a method and a system for recognizing dorsal hand vein based on in-bit-plane block mutual information. The method and the system generate eight bit-plane maps from gray-scale images retaining the dorsal hand vein contour, and extract the mutual information of the blocks in the bit-plane maps as features, so that the method has high robustness to image distortion and the like, high intra-class correlation and low inter-class correlation of the features, good separability, and high recognition rate.

The invention relates to a license plate recognition method based on deep learning, and belongs to the technical field of license plate recognition. The pictures (or videos) are collected through a front camera, then a convolutional neural network (CNN) is adopted to carry out feature extraction on the pictures so as to carry out target detection, and a multi-label classification method is adopted to carry out multi-label classification on license plates so as to obtain license plate characters. The license plate characters are identified by adopting a multi-label classification method, and the problem of low final identification rate caused by inaccurate segmentation in the license plate character segmentation process is avoided. The method has low requirements on the position of the camera, can be used in most occasions, and has wide application scenes.

The invention provides a face data processing method, a memory and a processor. The method is characterized by comprising the following steps: S11, inputting a face picture, and detecting feature dataof the face picture; S12, filtering a non-aligned face picture according to the feature data of the face picture; and S13, storing the aligned face picture. By filtering non-human faces, fuzzy humanfaces, side faces and light-colored human faces, the false recognition rate can be reduced to a large extent, the human face alignment method can be improved, human face alignment is achieved and sideface filtering is completed under the condition that operation is not increased, and therefore consumption of computing resources is reduced. The face recognition error recognition rate is reduced, the consumption of system computing resources is reduced, and the invention can be universally applied to the field of face recognition.

The invention discloses a West Lake Longjing traceability technology based on a new retail mode, belongs to the technical field of geographical indication product authenticity identification, and aims to solve the problems that single detection data cannot represent all key information of origin traceability and different types of detection data are combined in a metrology method for data matching and the like. According to the method, near-infrared characteristic spectrum, stable isotope, trace element and catechin data of Longjing wells in different producing areas are fused together based on a partial least square discrimination model, an analysis model is established, the Longjing well producing area is objectively and accurately determined by using the model after a sample is extracted, and the identification rate is the highest, reaches 100.0% and is higher than the PLSDA discrimination result of single data; the recognition rate for a blind sample is up to 100%, and the method has a good application prospect, and can be used as a West Lake Longjing production place traceability identification technical method.