153 results about "Computer vision and image processing" patented technology

A class of measurement devices can be made available using a family of projection patterns and image processing and computer vision algorithms. The proposed system involves a camera system, one or more structured light source, or a special pattern that is already drawn on the object under measurement. The camera system uses computer vision and image processing techniques to measure the real length of the projected pattern. The method can be extended to measure the volumes of boxes, or angles on planar surfaces.

The invention discloses an image significance detection method which uses confrontation training to generate a convolution neural network model, which belongs to the field of computer vision and image processing. The method comprises the steps of data preprocessing, network structure, suitable parameter selecting, and training with a random gradient descending method and an impulse unit. According to data preprocessing, a large amount of collected data and labels are preprocessed. According to network structure, a network structure and a specific kernel function are designed. Suitable parameters including learning rate, a momentum factor and the number of images inserted into the network each time are selected. The random gradient descending method and the impulse unit are used for training to reduce the possibility of network over-fitting. According to the invention, a significance map can be accurately acquired.

The invention belongs to the field of computer vision and image processing, and provides a rapid non-local super-resolution rebuilding method based on a minimum spanning tree (MST). Time complexity of an algorithm is reduced, meanwhile the quality of a rebuilding result is maintained, and a good balance between the time complexity and the quality is achieved. According to the technical scheme, the rapid non-local super-resolution rebuilding method based on the MST comprises the following steps: (1) obtaining an initial low-resolution depth image required to be rebuilt, then carrying out further simple pre-processing on the initial depth image, (2) obtaining a standard 4-connected planar graph, (3) obtaining the MST corresponding to an undirected connected graph in the step (2), and (4) rebuilding the coarse depth image after pre-processed according to the MST. The non-local super-resolution rebuilding method based on the MST is mainly applied in image processing.

The invention discloses a computer vision and radio frequency identification technology based book enrolment management apparatus, which completes the book enrolment flow in the one-apparatus, one-stop and one-step ways, and creates a new foundation for the future automatic book management requirement. The technical solution is that: the own characteristic of the book is extracted and analyzed by the embedded information technology, computer vision and image processing technology and Radio Frequency Identification (RFID) technology so as to fast collect, store, rectify, combine and analyze the image and acquire the characteristic of each registered book, and record them in the storage database and the RFID tag.

The invention relates to an image classification and searching method and provides the image classification and searching method based on geographic position characteristics and overall situation vision characteristics. The image classification and searching method aims at solving the problems that the precision is low when an existing outdoor image is subjected to scene classification by vision characteristics and a lot of time for an image searching process is spent along the gradual enlarging of a database scale, the instantaneity of a navigation positioning algorithm is influenced, and the position service requirements of a user cannot be guaranteed. The image classification and searching method is realized through the following technical scheme: step 1, initializing a database; step 2, pre-processing a database image; step 3, gathering the database image; step 4, sorting a database image type; and step 5, searching in the image type. The image classification and searching method is applied to the field of computer vision and image processing in an information technology.

The invention discloses a method for identifying tea leaves based on colors and shapes. By comprehensively utilizing a computer vision and image processing technology, increasing shape parameters of the tea leaves and improving a neural network algorithm, the automation for indentifying the qualify of the tea leaves is realized. The method comprises the following steps of: directly obtaining tea leaf images by a digital camera or a video camera and carrying out conversion and pre-treatment on image formats; and then, enabling tea leaf color characteristic parameters extracted based on an HSI (Hue-Saturation-Intensity) model and tea leaf shape characteristic parameters extracted by binary images to pass through a genetic neural network; and finally, finishing the automatic identification of the tea leaves. Through the way, a better identification effect can be obtained and the processing time of colored images is greatly shortened, thereby high coincidence between a detected result and a manual detected result is realized.

The invention discloses a method of using illumination estimation to correct an image color, which belongs to the technical field of computer vision and image processing. The method is quick in speed and good in robustness, and image color correction can be carried out in real time. The color is one of the most basic and the most direct features in vision information, and can be widely applied to application fields such as image retrieval and object recognition. However, the color feature is also a vision feature which is extremely instable, and is likely to be influenced by illumination changes. The image in an unknown illumination condition is corrected to an image under a standard white light, and the process is simply summarized that the illumination color in the case of image imaging is estimated firstly, and the image is mapped to the standard white light by using a Von Kries model. Better image white balance effects can be acquired. The method of the invention has the advantages that the parameters are few; the calculation is simple; the speed is quick; the effects are good; real-time processing can be carried out; and the method is particularly suitable for using a pretreated front end of a built-in camera to carry out color correction on the image.

The invention provides a method for controlling the distance between eyes of a user and the screen of electronic equipment in a computer vision and image processing mode. The method comprises the steps of face detection, pupil positioning, pupil distance computation and distance control. When the distance between the user and the screen of the electronic equipment is judged to be too short, prompting information is transmitted to require the user to adjust the distance between the eyes and the screen of the electronic equipment. Due to the application of the technology, the hardware cost of the electronic equipment cannot be increased, and the effect of protecting the vision of the user can also be achieved.

The invention relates to the computer vision and image processing technology, and proposes an autonomous positioning method for an unmanned aerial vehicle. An unmanned aerial vehicle autonomous positioning method based on visual SLAM (Simultaneous Localization and Mapping) is composed of a feature extraction and matching solution motion part, an image and inertia measurement unit imu fusion part and an estimation 3D point depth part; the method comprises the following steps: in the motion solving part, adopting a strategy for combining a feature point method and a direct method, selecting a key frame, after that, extracting point features and line features, and then minimizing an error to complete the calculation of the relative pose; in the image and inertia measurement unit imu fusion part, minimizing the error to fuse the error; and finally, in the estimation 3D point depth part, based on the matching of selected feature points, performing a triangulation method to solve 3D positions of the points, that is, depth values of the points are obtained. The unmanned aerial vehicle autonomous positioning method based on the visual SLAM provided by the invention is mainly applied to computer vision and image processing occasions.

The invention discloses a method for positioning facial features based on an improved ASM (Active Shape Model) algorithm, belonging to the technical field of visual and image treatment of a computer. The method comprises the following steps of: firstly, manually calibrating feature points; secondly, establishing a statistical shape model and a local gray model of an upper model and a lower model; thirdly, separately searching and matching the feature points in the upper model and the lower model; and finally, generating an example of a comprehensive model restrained by an energy function. To solve the problem that the traditional ASM method is difficult to position features under the facial expression conditions of a human face, the method for positioning facial features based on the improved ASM algorithm, disclosed by the invention, comprises the following steps of: carrying out the regional division of the facial features into an upper shape region and a lower shape region according to the change correlation, separately modeling the statistical shape model and the local gray model, generating examples of comprehensive shapes for the upper model and the lower model to restrain errors by introducing an energy function to the feature matching and searching process, and finally, obtaining an accurate feature positioning result. Due to the method for positioning facial features based on the improved ASM algorithm, the feature positioning accuracy of the ASM algorithm to the existing facial expression conditions are further improved.

The invention discloses a distributed cable external damage prevention system based on intelligent video detection. A distributed optical fiber is adopted to realize a sensing trigger function of preventing external damage in the whole cable laying process along with cable laying. A standby camera is awakened by adopting the trigger signal. A scene at a preset monitoring position is captured. Thescene picture is grabbed to a target detection server. A behavior subject of the intrusion event is identified and acquired on the server, and a judgment is made that the behavior hazard. The system can realize all-weather unattended operation, fully covers a cable laying line, realizes standby awakening through vibration triggering, has relatively low power consumption, and is suitable for photovoltaic power supply and other field scenes. Particularly, the system can automatically identify the category of an invasion object by utilizing a target detection technology related to computer visionand image processing, and an effective monitoring means is provided for the detection of the intelligent power grid.

The invention applies to protect a three-dimension locating method based on a three-point collineation marker in a video frame, and relates to the computer vision and image processing technology. The invention, aiming at the defects that the existing track locating algorithm based on plane markers is limited to plane objects and needs larger visual planes, provides a three-dimension locating method based on a three-point collineation marker, which is implemented by obtaining the two-dimensional coordinate information of the locating marker in the video frame, and establishing the relationship between the coordinate of the marker in a locating marker coordinate system and the coordinate of the marker in an image coordinate system to obtain the three-dimension location information and three-dimension posture information of the locating marker in a camera coordinate system. The invention is applied to three-dimension locating of an object having three-point collineation marker in the video frame, and provides a fast and precise tracking method for three-dimension space poin-and-click interactive operation.

The invention relates to the field of computer vision and image processing. The invention discloses a systematic design method for enhancing reality based on mobile platform. The method comprises: firstly, through a mobile device camera, previewing a human face in real time, and extracting image frames from a video streaming in real time, and meanwhile performing face detection and tracking on a signal frame image, to determine the position and size of the human face; on the basis, according to the real-time position of the human face, and rendering a three-dimensional scene in real time in a present video image, combining with the human face, according to the position of the human face, performing corresponding translation and zooming on a pattern, so as to achieve effect of enhancing reality. The system and the method can be applied in video application software in a mobile platform, and improve real-time and accuracy of face detection and tracking, and meanwhile, through rendering a virtual three-dimensional scene, effect of enhancing reality is achieved.

The invention provides a mobile phone playing behavior detection and recognition method based on human body posture estimation, and belongs to the field of computer vision and image processing. According to the technical scheme, the method comprises the steps that firstly, a monitoring system is initialized; secondly, the monitoring system collects data; secondly, the monitoring system monitors 2Dkey points of the human body and positions a framework; then, the monitoring system detects the mobile phone near the joint point hand; then, the monitoring system constructs a human body specific joint space pose relation; and finally, the monitoring system presets a mobile phone playing state and a mobile phone non-playing state according to the posture of the human body and whether the mobilephone is detected or not and the spatial position and posture relation between the mobile phone and the human body, and if the monitoring system judges that the human body detected in the current monitoring scene is in the mobile phone playing state, the monitoring system gives an alarm. According to the invention, an alarm prompt can be accurately sent out under the condition that a detected person plays a mobile phone, and meanwhile, the monitoring mode has the characteristics of non-contact, non-immersion, wide monitoring range, easiness in implementation and the like.

The invention belongs to the computer vision and image processing technology, and relates to a semi-supervised retina OCT image layer segmentation method combined with a generative adversarial network, which comprises the following steps: preparing retina OCT image data, taking labeled pictures of part of patients and unlabeled pictures of all patients as a training set, and taking labeled pictures of other patients as a test set; constructing a generative adversarial network, the generative adversarial network comprising a segmentation network and a discriminator network, and the output end of the segmentation network being connected with the input end of the discriminator network; designing a loss function of the generative adversarial network; setting evaluation indexes; and introducingthe prepared training set by using the designed loss function, and training the generative adversarial network. According to the method, the generative adversarial network is trained by utilizing thelabeled data and the unlabeled data at the same time, so that the robustness of the network is enhanced, and the semantic segmentation accuracy is improved.

The invention provides an indoor positioning method based on visual feature matching and shooting angle estimation. The invention relates to the indoor positioning method based on visual feature matching and shooting angle estimation, so as to solve the problems of a lot of early equipment and infrastructure investment and stringent scene positioning requirement of the existing indoor positioning method. According to the technical scheme, a visual database in an indoor scene is built, wherein the visual database comprises the position coordinate of a visual feature in an indoor coordinate system and the camera coordinate system of the visual feature in an indoor acquisition process; according to the visual database acquired in the first step in the indoor scene, a database image which matches query images P1Q and P2Q to the greatest extent is solved; a rotation matrix R1 corresponding to the query image P1Q and a rotation matrix R2 corresponding to the query image P2Q are calculated; and according to the R1 and the R2, the position of a query image camera is calculated. The indoor positioning method provided by the invention is applied to the field of computer vision and image processing.

The invention is applicable to the technical field of computer vision and image processing, and provides a lane line detection method and device, terminal equipment and a readable storage medium. Inputting the road image into a trained neural network model for processing, and outputting a detection result of a lane line in the road image of the current scene; wherein the trained neural network model is obtained by training according to sample images in a training set and a semantic segmentation model, and the sample images in the training set comprise collected road images of a plurality of scenes and marked images corresponding to the road images of the plurality of scenes. According to the method and the device, the problems that most of deep learning models used for lane recognition atpresent are relatively large in calculation amount, relatively complex in model and unfavorable for meeting the requirement on real-time performance in an actual application scene of an automatic driving task can be solved.

The invention discloses an automatic making up method of a three-dimensional Beijing opera facial makeup. The method comprises steps of acquiring image input from an input source, carrying out face detection and face feature point positioning and marking the feature points on an image; carrying out face normalization processing on the feature points, calculating depth information of the normalized face image and according to the depth information, reconstructing three-dimensional point cloud; according to the feature point information, carrying out triangular blocking operation on the face image; manually marking the feature points on a prepared Beijing opera facial makeup picture, and carrying out the triangular blocking operation on the Beijing opera facial makeup; coloring the face image with each triangular block on the Beijing opera facial makeup; and using the Beijing opera facial makeup to recolor the three-dimensional point cloud, and displaying the result to a user. According to the invention, by calculating the coordinates of the face feature points and the face three-dimensional mode, and using the technology related to computer vision and image processing, a Beijing opera facial makeup with manually pre-marked feature points is drawn on the three-dimensional model of the target face.

The invention is suitable for the technical field of computer vision and image processing, and discloses a video deblurring method, device and equipment based on ambiguity, and the method comprises the steps: calculating the ambiguity of a video frame; determining a clear frame and a fuzzy frame according to the ambiguity; generating a reference frame according to the clear frame and the fuzzy frame; performing image block extraction on the fuzzy frame and the reference frame; performing weighted fusion according to the weights corresponding to the pixel points in the image block to obtain a fused image block; and recombining the fused image blocks to obtain an output image. According to the embodiment of the invention, the ambiguity of the video frame is calculated without estimating theambiguity kernel, and the clear frame and the ambiguity frame are determined according to the ambiguity, so that the calculation complexity is effectively reduced, and the calculation speed is increased; and the weight of the reference frame is considered, and weighted fusion is carried out according to the weight corresponding to the pixel points in the extracted image block, so that the definition of the finally output image is relatively high.

The invention discloses a human body target part automatic analytic method based on multiple images, and belongs to the field of computer vision and image processing. According to the method, on the basis that the multiple human body target images are given, by means of concurrence among targets, human body target parts in the images are automatically analyzed. By means of the method, a human body target classifying module and a human body target analyzing module are executed in an iterated mode until the stopping condition is reached. The human body target classifying module achieves automatic classification of the multiple human body target images. In a classifying process, appearance similarity and gesture similarity among the human body targets are considered comprehensively; the human body target analyzing module achieves automatic analysis of the human body target parts in each classification and alignment of analyzed results of different classifications. According to the method, independent analysis of each human body target image is converted into simultaneous analysis of the multiple human body target images, and the limitation caused by a traditional analysis mode that learning is conducted and then inference is conducted is overcome.

The invention belongs to the technical field of computer vision and image processing and particularly relates to a method for correcting image colors through gray points. The method comprises dividing an image into three components of red, green and blue and performing logarithm transformation on each channel; calculating the local contrast of each logarithm channel and calculating the average gray value obtained by dividing the contrast ratio relative standard deviation of all pixel points in the three channels by the pixel points in the positions to obtain a gray measurement coefficient distribution diagram; sorting all pixels of the gray measurement coefficient distribution diagram from the small gray measurement coefficient to the large gray measurement coefficient, extracting the top 50% of the pixel points of the total pixel points to serve as gray points to estimate the image light source, and performing color correction on the image by using the light source. The method has the advantages of being few in parameter, simple in calculation, rapid in speed, good in effect, capable of performing real-time processing and the like and particularly applicable to color correction of images through front ends arranged in camera preprocessing units.

The invention discloses a household waste sorting method based on machine vision. The method comprises: step 1, performing constant volume treatment on wastes to be sorted; step 2, forwardly delivering the wastes subjected to constant volume treatment with a delivery belt, sorting magnetic metal wastes using a metal waste sorting device, and putting the magnetic metal wastes into a magnetic wastecollecting device using a mechanical arm sorting device; step 3, continuously photographing the wastes moving at constant speed, feeding the photographed waste image to a database to be compared withand identify inherent image in the database, thereby judging whether the waste is non-fruit and vegetable waste; step 4, calculating weighted mean of geometric positions of all the outline pixels to obtain the position coordinates of the target waste in the delivery belt, and transmitting this position information to a control system. The present application utilizes a computer vision and image processing technology to sort non-fruit and vegetable wastes such as metal, bottles, PVC and stones in household waste, and fruit and vegetable wastes are fed to the next process.

The embodiment of the invention discloses a gaze estimation method and a gaze estimation device applied to a scene at least including double glints, relates to the field of computer vision and image processing, and is used for improving the adaptability to the head movement and improving the precision and the stability of the gaze estimation. The method provided by the embodiment of the invention comprises the following steps: obtaining an eye appearance descriptor and a PCCR (Pupil Center-Cornea Reflection) descriptor based on the double glints; carrying out feature fusion on the glints-centered descriptor and the PCCR descriptor to obtain a mixed gaze feature descriptor, wherein the mixed gaze feature descriptor comprises a mixed gaze feature descriptor of a training sample and a mixed gaze feature descriptor of a test sample; and carrying out PoR (Point of Regard) estimation on the test sample according to the mixed gaze feature descriptor of the training sample to obtain the PoR coordinate of the test sample.

The invention provides a navigation positioning device and method in a spacecraft cluster ground simulation system, and belongs to the technical field of computer vision and image processing. According to the device, a red optical filter is arranged on a lens of an industrial camera in the positioning device; a camera bracket is fixedly connected with the industrial camera and used for supportingthe industrial camera; a moving target performs one-dimensional rotation and two-dimensional translation on a test table; the industrial camera collects an image containing the moving target and transmits the image to a pose measurement PC machine; the pose measurement PC machine calculates the pose information of the moving target in a three-dimensional space according to the two-dimensional coordinates of the image; and a positioning method comprises the steps of camera calibration, image collection, image preprocessing, contour extraction, centroid calculation, pose calculation and the like. According to the invention, multi-target pose measurement can be realized; the operation speed is increased, and the requirement for real-time performance in actual engineering can be met; and the interference in a complex environment can be effectively suppressed, and the anti-interference capability is improved.