45results about How to "Improve image recognition rate" patented technology

Extensive social communication is induced. Connection is made with a network terminal capable of connecting to the Internet, and an image and voice signal reflecting the subjective visual field of the user and the like which can be obtained from the headset system that can be worn by the user on the head is uploaded via the network terminal to a knowledge-information-processing server system, and specifying and selecting of an attention-given target by the voice of the user himself/herself are enabled on the server system with collaborative operation with the voice recognition system with regard to a specific object and the like to which the user gives attention and which is included in the image, and with regard to the series of image recognition processes and image recognition result made by the user, image recognition result and recognition processes thereof are notified as voice information to an earphone incorporated into the headset system of the user by way of the user's network terminal via the Internet by the server system with collaborative operation with a voice-synthesizing system, so that user's message or tweet can be extensively shared by users.

The invention discloses a space positioning device, system and method. The space positioning device comprises a communication unit for establishing communication with a positioning object; an image obtaining unit for obtaining a first stereo image of a background image under the condition that a mark point infrared light source of the positioning object is in an off state, and a second stereo image comprising a mark point under the condition that the mark point infrared light source is in an on state; an image processing unit for carrying out image processing on the first stereo image and the second stereo image to obtain a mark point light spot graph; and a positioning unit for obtaining a space coordinate of the mark point according to the mark point light spot graph. The space positioning device, system and method realize distinguishing between the object to be positioned and the background image, improve image recognition rate, reduce algorithm complexity, simplify algorithm operation environment and save cost.

The invention discloses an automatic inspection and real-time image acquisition transmission method of a railway line unmanned aerial plane. Through the method, automatic track plan inspection can be realized, visual navigation, image identification and obstacle avoidance are supported, parameters can be properly changed as no transmission delay influence is exerted, so the image identification rate is improved, a satellite communication network mode is fused with a traditional communication mode, high speed exchange for high-capacity image data can be realized, and the method further has an advantage of relatively high security.

The present invention relates to a food material input method and a food material input device for a refrigerator. The food material input method comprises: obtaining a food material image inside a storage room; identifying N kinds of predicted food materials and the confidence thereof according to the food material image, wherein N is a positive integer; and determining the type of the food material according to the N kinds of predicted food materials and the confidence thereof. In addition, the present invention also provides a food material input device by using the method. By virtue of thetechnical scheme of the present invention, in the food identification process, the accuracy of identification can be improved, the waste of resources can be reduced, and the user experience can be improved.

The invention relates to a feature weighted face identification algorithm, belongs to the field of human face technology. The feature weighted face identification algorithm comprises the steps that a face image is decomposed into high and low frequency components via wavelet transformation, principal component analysis (PCA) is carried out on the different components to extract feature images, weighing is carried out by an AHP algorithm according to the significance of the components, and a support vector machine (SVM) is used for classification and identification. The image identification rate is highest, and all main information of images is used for identification.

A vehicle safety control apparatus using cameras includes a first camera configured to photograph a current object at at least one of a narrow angle and a straight angle during running, a second camera configured to photograph the current object at a wide angle during the running, an image processing unit configured to perform image processing on first current object image data captured by the first camera and second current object image data captured by the second camera, a recognition unit configured to recognize the first and second current object image data on which the image processing unit has performed the image processing, a storage unit configured to cause the recognized data to match preset reference object-specific data and separately store the data matching the preset reference object-specific data; and a control unit configured to receive the recognized data and deliver a storage command to the storage unit.

The embodiment of the present invention provides a method and device for protecting a camera against fog interference, and an electric appliance, and belongs to the field of electric appliances. The method includes: determining the lens temperature of the lens of the camera; determining the dew point temperature at the lens; and instructing a heating module to heat the lens according to the dew point temperature at the lens and the lens temperature. The device includes: a heating module for heating the lens of the camera; a lens temperature determination module for determining the lens temperature of the lens; a dew point temperature determination module for determining the dew point temperature at the lens; and a control module for instructing the heating module to heat the lens accordingto the dew point temperature at the lens and the lens temperature. Therefore, no fog is generated on the surface of the camera, or the fog is quickly removed after generated, thereby protecting the camera against fog interference and reducing the influence of the fog on the imaging effect of the camera.

The invention provides a transformation-based X-ray image BGA (ball grid array) solder ball contour extraction method. The method includes steps: S1, acquiring a solder joint original image, subjecting the original image to circle detection to obtain a solder joint image, and subjecting solder balls of each solder joint to primary recognition to obtain ROI (region of interest) of each solder joint; S2, transforming the original image to obtain a target image, and subjecting the target image to edge detection; S3, mapping a detection result to an original coordinate system to obtain a mask image, and subjecting the original image to mask processing to extract images of single solder joints; subjecting the mask image to edge extraction to obtain a solder joint contour image.

The invention belongs to the technical field of image recognition, and particularly relates to an image recognition method based on a convolutional neural network. According to the method, different segmentation methods are adopted to perform image segmentation on the same image; and for different segmentation methods, different feature extraction algorithms are adopted to extract image features,the extracted image features are inputted to the same convolutional neural network model for identification, and the final identification result of the image is calculated in combination with the identification results of different weights.

The invention discloses a tree species image recognition method based on a natural background. The method comprises the steps of firstly capturing tree image data through employing a web crawler technology, storing the tree image data according to different types, and forming a data set; preprocessing the image, and segmenting a complex background in the original image from a target tree entity byadopting a Grab-Cut image segmentation method; learning the segmented image by using a convolutional neural network; adopting k-fold cross validation for model selection, wherein firstly, known dataare randomly segmented into K subsets which are not intersected with one another and are the same in size; training a model by using the data of K-1 subsets, and testing the model by using the remaining subsets; repeating the process for K times, finally selecting the model with the minimum average test error in K times of evaluation, and identifying the tree species image. According to the method, an image segmentation method is adopted, complex backgrounds which are not used for tree species identification are segmented, and the tree species identification rate is greatly improved.

An apparatus and methods for recognizing an image with an increased image recognition rate are provided. The apparatus includes a plurality of illuminators for illuminating a surface. The apparatus also includes an image sensor for outputting an image frame obtained by converting an optical image of the surface into an electrical signal. The surface is illuminated one of the plurality of illuminators. The apparatus further includes a switch for selecting any of the plurality of illuminators for illuminating the surface. The apparatus additionally includes a controller for receiving the image frame from the image sensor, determining whether the image frame has a region of reflected light, and controlling the switch to select another of the plurality of illuminators to illuminate the surface when the image frame has the region of reflected light.

The invention relates to a color image recognition method which is characterized by comprising the following steps: S1, acquiring a picture of a processing object; s2, performing channel selection processing; s3, executing subsequent processing for the selected specific channel; s4, extracting characteristic values of the characters according to a set recognition algorithm after processing is completed, comparing the characteristic values with a target value of the model, and calculating and judging a closest result; and S5, outputting identification result information. The method has the advantages that after the method is used, the original color information of the image can be utilized to the maximum extent, the contrast ratio and the discrimination degree are increased, information loss caused by gray processing is avoided, the recognition effect of the feature value is improved under the condition that other feature value recognition steps are not changed, and finally the image recognition rate is increased.

The invention discloses a man-machine voice interaction system and method for a shopping bin, and relates to the technical field of intelligent interaction. The method comprises the steps of samplingdata; acquiring multi-dimensional information by using user voice and a user identifier, and determining an identification model; allowing a user to collect a front-side image through photographing equipment; detecting the acquired image through face detection equipment, and carrying out human eye positioning; intercepting a face area in the image, and performing graying processing; storing the processed data; extracting SIFT features in the image to serve as training sample data to be stored in a data storage center; dynamically updating identification resources according to the division; anddetermining different user information of going in and out of the shopping bin by using a user voice recognition text. According to the man-machine voice interaction method for the shopping bin, thefunctions are diversified, the working efficiency is high, the maintenance cost is low, a new module can be added in later maintenance and updating to meet the requirements of a later system, and thefunctions of the man-machine voice interaction method for the shopping bin are more complete.

The invention discloses an automatic building drawing information extraction method and system based on an image recognition technology, and the method comprises the steps: carrying out manual marking of a matching anchor point and a keyword character region of each building drawing template in advance, and obtaining a type matching template of a matching type template library; identifying the type matching template with the highest matching degree, and finding a plurality of component sub-matching templates matching the type template; extracting contour information of a to-be-recognized building drawing, performing component matching when each piece of contour information matches the sub-matching template of the component, and identifying component information of the to-be-recognized building drawing; or comparing each component sub-matching template in the component sub-matching templates with the contour of the architectural drawing to be identified so as to find the component information of the to-be-recognized architectural drawing; and intercepting the component information of the to-be-recognized architectural drawing, identifying the non-components of the to-be-recognized architectural drawing by using an SSD algorithm, and forming digital information of the to-be-recognized architectural drawing.