56results about How to "Improve image processing speed" patented technology

The invention relates to a hybrid image processing method of extracting the centric line of the structured-light stripe based on the ROI area, belonging to the field of machine vision technique. By combining the image threshold processing, communicated components extraction, shape factor constraint and expansion method, the invention automatically divides the structured light area as the ROI area of light stripe extraction, using the extraction method of fringe center based on Hessian matrix to realize the extraction of sub-pixel image space of the centric line of the structured-light stripe of the ROI area. The method has the advantages of high precision, good robustness, high automation degree and other advantages, avoiding the large formwork Gaussian convolution calculation of the pixel the non-light stripe area, thereby reducing the redundant calculation of the extraction of the structured light stripe and realizing the high speed and precision extraction of the centric line of the light stripe, laying a foundation for the real-time application of the structured light 3D vision measurement.

A solid-state image device is provided which includes: a photoelectric conversion portion which obtains a signal charge by photoelectric conversion of incident light; a pixel transistor portion which outputs a signal charge generated by the photoelectric conversion portion; a peripheral circuit portion which is provided at the periphery of a pixel portion including the photoelectric conversion portion and the pixel transistor portion and which has an NMOS transistor and a PMOS transistor; a first stress liner film which has a compressive stress and which is provided on the PMOS transistor; and a second stress liner film which has a tensile stress and which is provided on the NMOS transistor. In the solid-state image device described above, the photoelectric conversion portion, the pixel transistor portion, and the peripheral circuit portion are provided in and/or on a semiconductor substrate.

The invention provides an industrial intelligent camera which comprises an imaging module, a master control module, a transmission module, a power module and a storage module. The imaging module comprises a CMOS image sensor and a peripheral circuit; the master control module is an FPGA chip and comprises an image collecting control module, an image pre-processing module, an image advanced processing module, a transmission control module and a storage control module. According to the industrial intelligent camera, all image processing functions are achieved through the FPGA chip, the complex interactive design of an FPGA and a DSP is avoided, and the advantages that algorithms of the FPGA is achieved through a hardware circuit, so that the processing speed is high, and performance is stable are sufficiently played; the industrial intelligent camera is high in integration degree, small in overall size, small in weight, less in power dissipation and particularly suitable for application occasions with high requirements for the camera size and the image processing speed.

The invention relates to a remote control interactive pen. The remote control interactive pen comprises a power supply, a laser module, a control module and a lens group, wherein a reflector is arranged in the lens group; the laser module comprises a visible laser module and an infrared laser module; visible light emitted from the visible laser module and infrared light emitted from the infrared laser module are reflected by the lens group and are emergent along the same optical axis; the visible laser module emits the visible light to enable an operator to accurately know the position and the moving locus of the interactive pen and freely control the movement of the interactive pen; the infrared laser module emits the infrared light; and a receiver detects the projection position of the infrared light and converts the position information into a mouse signal for the interactive operation. The invention further provides the receiver of the remote control interactive pen, wherein the receiver is provided with a narrow band pass filter which corresponds to the spectrum of the infrared light emitted from the interactive pen. The receiver has the advantages of reducing the interference of outer light and improving the work stability.

Embodiments of the invention provide an image processing method, apparatus and system. The method comprises the steps that a central processing unit (CPU) writes first image data corresponding to a first image in a first buffer region; and the CPU writes second image data corresponding to a second image in a second buffer region, so that an image processor can conveniently perform superposition processing on the first image and the second image based on the first image data and the second image data. According to the image processing method, apparatus and system provided by the embodiments of the invention, the image processing can be finished only by two buffer regions while a UI thread and a browser engine are ensured to be independently refreshed, so that the memory consumption is reduced; and in addition, the image data is processed by the image processor instead of the CPU, so that the transmission time of the data on a bus is shortened, the image processing speed is increased, and the CPU burden is reduced.

The invention discloses a welding device based on the binocular vision sensing of structured light. The welding device comprises an installation bracket, a structured light emitter, a data storage unit, a binocular vision system, an automatic welding device and a motion control system, wherein the structured light emitter is used for positioning the center of a pipe plate pattern to be welded and movably arranged on the installation bracket; the data storage unit is used for storing the position information of the pipe plate pattern to be welded, which is obtained by the structured light emitter; the binocular vision system is provided with a first vision device and a second vision device which are movably arranged on two sides of the structured light emitter; and the motion control system is used for controlling the three-degree-of-freedom motion of the automatic welding device during welding of the pipe plate pattern to be welded. The welding device based on the binocular vision sensing of the structured light has the advantages that by using the binocular vision system and adopting a spatial circle identification algorithm, the center of the pipe plate pattern to be welded in a certain inclination angle range can be identified; the characteristic points can be quickly and accurately identified; image processing speed is increased; and the positioning speed of a system pipe hole is greatly increased.

The invention provides a nuclear rod pose automatic identification device and method and belongs to the technical field of machine vision and nuclear power equipment manufacturing automation. The device comprises a first camera and a second camera which are orthogonally arranged. The optical axis of the first camera is parallel to a Z axis, and horizontal and vertical coordinates of an image are parallel to an X axis and a Y axis respectively. The optical axis of the second camera is parallel to the X axis, and horizontal and vertical coordinates of an image are parallel to the Y axis and the Z axis respectively. The method comprises steps that firstly, image features of a lower plug of a nuclear rod are extracted, then space pose information of the lower plug of the nuclear rod is calculated, and finally the nuclear rod is assembled according to the space pose information. The automatic identification device and method can monitor shape change and pose change of the nuclear rod, provide a device basis and a theoretical basis for automatic assembly of the nuclear rod and a grid plate, and is beneficial for achieving automatic assembly of nuclear rod clusters. The safety of nuclear power equipment manufacturing is improved, and the intelligent level of nuclear power equipment manufacturing is greatly improved.

The invention belongs to the field of security and protection and the technical field based on biological recognition, and particularly relates to an intelligent refrigerator with an iris recognition function. The intelligent refrigerator comprises a refrigerator body, an iris gathering module, a central control module, a lock module, an alarm module, a man-machine interaction module, a communication interface module and a power circuit module. All the above modules constitute an intelligent security controller, so that an active anti-theft and intelligent anti-theft unlocking scheme in a non-contact mode is realized. According to the intelligent refrigerator with the iris recognition function, an industrial host is used in the central control module, the dominant frequency can reach 1.8 GHz, and thus the image processing and storing speed is increased. User experience is improved through a display screen with a LCD under the condition that security is guaranteed. The security of the refrigerator is improved based on an alarm mode that short messages are sent to mobile phones in real time. The user identity uniqueness and reliability of the refrigerator are improved based on a non-contact identity recognition verification door opening mode of a binoculus iris image.

The invention discloses a self-adaptive vision lane departure pre-warning device and relates to a vehicle safety system. The self-adaptive vision lane departure pre-warning device is provided with a vehicle-mounted power supply, a voltage-reducing and voltage-stabilizing circuit, an alerter, an FPGA (Field Programmable Gate Array) core board, a digital camera and a steering angle inductor, wherein the input end of the voltage-reducing and voltage-stabilizing circuit is connected with the output end of the vehicle-mounted power supply, the output end of the voltage-reducing and voltage-stabilizing circuit is respectively connected with the input end of a power module of the FPGA core board and the input end of the digital camera, an output interface of the FPGA core board is connected with the input end of the alerter, the output end of the digital camera is connected with an input interface of the FPGA core board, the output end of the digital camera is connected with the input interface of the FPGA core board, and the output end of the steering angle inductor is connected with the input interface of the FPGA core board. The lane line identification capability of the self-adaptive vision lane departure pre-warning device under the conditions of complicated illumination and complicated road surfaces is improved. The departure pre-warning capability of the vision lane departure pre-warning device is improved. The image processing speed and the pre-warning response speed of the traditional vision lane departure pre-warning device are improved.

The invention discloses a zoom multi-channel microscopic imaging system of eye retina, which comprises a lighting subsystem, a zooming module, a multi-channel imaging module and a controller, wherein the lighting subsystem is used for emitting light to light eye ground, and the light is reflected by retina tissues and then forms imaging light beams; the zooming module is suitable for optical zooming so as to focus-image all the depth zones of the retina and lead to different magnifying powers and fields of view; the multi-channel imaging module is used for filtering and arranging the imaging beams spatially after the imaging beams are split and finally focusing the light on the target surface of a photoelectric detector to form a plurality of twin images; and the controller is used for connecting and controlling all the other parts, taking pictures of the eye ground of the tested person continuously, and collecting and obtaining a video or an image sequence directly. The zoom multi-channel microscopic imaging system also comprises a self-adaptive optical subsystem which is used for correcting higher order aberration of human eyes in real time so as to reach an optical resolution ratio approaching the diffraction limit and realize microscopic observation of the living eye retina.

Provided are an optical system for processing an image using a PSF and an image processing method thereof. The optical system includes one or more lenses, an image sensor, and an image processor. The image sensor senses an image formed by the lenses, and the image processor recovers the image sensed by the image sensor using a PSF. A BFL defined as a distance between a refractive surface of one of the lenses that is closest to an image-side and the image sensor, is controlled such that ΔPSF defined by Equation below has a value less than 100%. $\Delta \mathrm{PSF}\left(\%\right)=\frac{{\mathrm{PSF}}_{\infty }-{\mathrm{PSF}}_{\mathrm{macro}}}{\left({\mathrm{PSF}}_{\infty }+{\mathrm{PSF}}_{\mathrm{macro}}\right)/2}\times 100$

An image is recovered using a PSF, so that an excellent image can be obtained over a wide object distance range.

The invention belongs to the technical field of field crop growth information processing, and discloses a field video image real-time segmentation system and method based on an FPGA, and the method comprises the steps: carrying out the change of the structure of an image segmentation algorithm suitable for a hardware environment, adding data and an assembly line instruction, and enabling the imagesegmentation algorithm to adapt to an application scene of real-time image segmentation; designing each module of an image acquisition and display system, and establishing a video path based on a camera; and achieving field video image segmentation in combination with processes of field plant ultra-green feature extraction, a Canny edge detection algorithm, median filtering and threshold segmentation and FPGA hardware development. According to the system and method, rich register resources and pipeline structures are provided, a large amount of image data can be processed in real time, and the image processing speed is greatly improved; the development board used in the invention is lower in cost, small in size, low in power consumption and more suitable for complex field working environments.

The invention discloses a method and device for determining parameters in sight tracking equipment. The method comprises the following steps of: obtaining a to-be-detected image and a feature matrix of the to-be-detected image; processing the to-be-detected image on the basis of a preset model so as to obtain a first vector and a second vector of the to-be-detected image; and determining parameters of the to-be-detected image on the basis of the first vector and the second vector, wherein the parameters of the to-be-detected image comprises at least one of positions of light spots in the to-be-detected image and sight directions of eyes in the to-be-detected image. According to the method and device, the technical problem that the sight estimation algorithms and light spot prediction algorithm of sight tracking equipment in the prior art are complex in calculation is solved.

Techniques are described for processing computerized images with a graphics processing unit (GPU) using an extended vertex cache. The techniques include creating an extended vertex cache coupled to a GPU pipeline to reduce an amount of data passing through the GPU pipeline. The GPU pipeline receives an image geometry for an image, and stores attributes for vertices within the image geometry in the extended vertex cache. The GPU pipeline only passes vertex coordinates that identify the vertices and vertex cache index values that indicate storage locations of the attributes for each of the vertices in the extended vertex cache to other processing stages along the GPU pipeline. The techniques described herein defer the setup of attribute gradients to just before attribute interpolation in the GPU pipeline. The vertex attributes may be retrieved from the extended vertex cache for attribute gradient setup just before attribute interpolation in the GPU pipeline.

The invention discloses a method for extracting image vein characteristics of a finger vein identification system, which converts the extraction of vein characteristics into the problem of detecting a valley shaped area in an image. The method comprises the following steps: using any one pixel point in a vein image as a seed to start growing, wherein a growing point performs growth of pixels one by one along each direction of the pixels; if the growing point is judged as a pixel point on the vein, growing continuously in the direction, otherwise, growing the growing point in other direction of a seed point; selecting points at each different position from the image randomly as the seed points to start growing; and finally, recording the total times of each tracked pixel point in a corresponding position in a track matrix, wherein the matrix has the same size as the image, so that the finger vein characteristics are obtained by selecting a threshold value to divide the track matrix. The method can accurately extract the vein characteristics in the image and reduce the extraction of error characteristics and the missing of real characteristics, and can improve the identification accuracy of the whole system and improve the speed of processing the image at the same time.

The invention discloses a method for automatic segmentation of a defective image. The method comprises the steps that A, image data are clustered through two-layer subtraction clustering; B, arrangement is conducted on center sets from big to small; C, a central point set V is initialized through first c elements; D, an efficiency analysis index of clustering is calculated; E, c=c+1 and an iterative computation is conducted until c>cmax; F, the value of V is determined when FXB (U,V,c) is the least value; G, U is calculated again through the value of V and image segmentation is conducted according to the formula that uik=max{u1k, u2k,...,uck} and xi belongs to the ith type. The method can overcome the shortages of the prior art, automatic segmentation in random walking mode is achieved, the image processing quality is improved, operation time is shortened, and the work efficiency is improved.

The invention discloses a laser positioning lamp cross curve image detection method. The method includes the steps that laser positioning lamp cross curve image data are acquired through a photoelectric sensor array; the cross curve image data are binarized; binarized images are thinned, and a set (phc) of transverse points of a thinned cross curve and a set (pvc) of longitudinal points of the thinned cross curve are acquired; the slope K of a straight line formed by any two points in the set (phc) of the transverse points and the slope K of a straight line formed by any two points in the set (pvc) of the longitudinal points are calculated, the slopes K of all straight lines are counted, and the slope K1 and the slope K2 which are highest in frequency of occurrence are obtained; the slope K1 and the slope K2 are substituted into the formula k1=tan alpha and the formula k2=tan beta to calculate the horizontal angle alpha and the vertical angle beta of the laser positioning lamp cross curve. The invention further discloses a laser positioning lamp cross curve image detection device. The laser positioning lamp cross curve image detection method has the advantages that detection accuracy, antijamming capability and the detection speed are high.

The invention discloses an enhanced CT image tracheal wall enhancement method, a system, a device and a medium, and the method comprises the steps: obtaining an enhanced CT image sequence, and carrying out the thresholding of the enhanced CT image sequence; performing three-dimensional region growth on the image subjected to thresholding processing to obtain lung and trachea masks; performing closed operation on lung and tracheal masks; performing three-dimensional region growth on the image obtained by the closed operation, and segmenting a trachea main body; calculating boundary pixel characteristics of each pixel point of the trachea main body; judging whether each pixel point of each image in the enhanced CT image sequence belongs to a tracheal wall based on the boundary pixel characteristics; and enhancing the pixel values of the pixel points belonging to the tracheal wall.

The invention discloses a disaster information rapid extraction method based on video image analysis, and belongs to the field of big data and artificial intelligence combined with geographic information. The disaster information rapid extraction method comprises the following steps: firstly, acquiring disaster area information, taking event occurrence GPS position data as a center, defining an influence area by using an intensity circle for rasterization, and screening out and grouping each population density area; then extracting cameras in each group, and dividing the cameras containing a certain architectural sign into effective camera groups; finally, for a certain effective camera, extracting key frames of time periods before and after the vibration time point T, respectively establishing key frame sequences, and respectively performing similarity calculation for comparison and detection; and similarly, calculating similarity comparison difference values of all effective cameras; and weighting all comparison difference values to obtain a damage index of the symbolic building, completing damage index labeling of each group of population density areas, and drawing a damage index contour line to form identification and labeling of a key target area. The disaster information rapid extraction method improves the image processing speed, and is higher in the range precision of the data set.

The invention belongs to the technical field of industrial welding, and particularly relates to an automatic plane correction method for an intelligent tube plate welding robot based on a cross laser. The method comprises the steps of sensor calibration, system hand-eye calibration, light strip center extraction and robot attitude angle deviation calculation. According to the method, the cross laser is directly projected to a tube plate plane to obtain three-dimensional data without being affected by the placement angle of the tube plate plane, so that the degree of freedom is larger; by adopting a gradual iteration approximation mode, the leveling precision can be well guaranteed, and the manual participation degree is greatly reduced; by means of iterative leveling, an ICP iterative algorithm based on the least square method is adopted in each leveling calculation, calculation results are converted into Euler angles, higher calculation precision is achieved, the leveling precision angle error is about 0.1 degree, and the method is suitable for most robot controllers and has high generalization performance.

An embodiment of the invention provides an image display method and device, relates to the field of data processing and is capable of solving the problems about image refresh display line by line and image retention due to the fact that copying speed of image data is too low. The image display method includes: acquiring the image data of an image to be displayed, acquiring a superposition area of the image to be displayed on a currently-displayed image when determining that the image to be displayed need to be superposed to display on the currently-displayed image on a user interface and establishing an image exchanging area identical to the image to be displayed in size; integrally copying the image data of the superposition area of the currently-displayed image to the image exchanging area; integrally storing the image data of the image to be displayed to a data storage area corresponding to the superposition area so as to display the image to be displayed on the superposition area; integrally copying the image data stored in the image exchanging area to the data storage area corresponding to the superposition area so as to redisplay the image corresponding to the image to be displayed in the superposition area. The image display method and device is applied to image display.

The invention relates to an in-vehicle article monitoring method, electronic equipment and a storage medium, relates to the technical field of Internet of Things, and is used for solving the problem that when a comprehensive upgrading mode is adopted during software upgrading, the influence range is relatively large when upgraded software runs abnormally, and the in-vehicle article monitoring method comprises: obtaining a first in-vehicle image of the target object before getting on the vehicle, a second in-vehicle image of the target object after leaving the vehicle, and at least one third in-vehicle image of the target object in the riding process; searching a first area with the similarity with the first in-vehicle image smaller than a first preset value from the second in-vehicle image; extracting a second region having the same image position as the first region from the third in-vehicle image; and if the similarity between the first area and the second area is greater than a second preset value, determining that the object of the target object is left in the vehicle. According to the embodiment of the invention, the possibly left article in the image is identified, and the possibly left article and the scene when the article is on the vehicle are verified again, so that the accuracy is improved.

The invention provides an imaging method and device applied to cone beam CT sparse sampling, a medium and a program product. The imaging method applied to cone beam CT sparse sampling comprises the following steps: determining a plurality of current imaging targets; splitting the original projection data into a plurality of original projection sub-data according to a plurality of current sampling rate requirements respectively corresponding to the plurality of current imaging targets; and performing imaging corresponding to each current imaging target in the plurality of current imaging targets according to the plurality of original projection sub-data by using a deep neural network model. Compared with an existing traditional scheme which can cause the problem of error amplification through a cyclic interpolation mode, the method can directly amplify subdivision position projection, meanwhile, a residual error and structural similarity double-constraint neural network model is combined, the fidelity of an amplified projection image is enhanced, the imaging speed and the image processing speed are greatly improved, and the image processing efficiency is improved. And under the condition of meeting low-dose irradiation, the signal-to-noise ratio, the structural similarity and the overall image quality of the reconstructed image are remarkably improved.

The invention relates to a household appliance control technology, and discloses a control method and device of a range hood, the range hood and a readable storage medium. The control method of the range hood comprises the following steps that image data of the range hood are acquired, and whether a user leaves or not is detected based on the image data; when that the user leaves is detected, theleaving duration of the user is acquired; whether the leaving duration exceeds a preset threshold value or not is judged; and when the leaving duration exceeds the preset threshold value, the range hood is turned off. According to the control method and device of the range hood, the range hood and the readable storage medium, whether the current range hood is in a user use state or not can be moreaccurately identified, when that the current range hood is not used by the user is judged, the range hood is turned off, so that electric energy is saved, the electric energy consumption of the rangehood is reduced, and potential safety hazards are avoided.