442results about How to "Avoid error accumulation" patented technology

The invention relates to a calibration technology for multiple structured light projected three-dimensional profile measuring heads, discloses a technology for calibrating multiple structured light projected three-dimensional profile measuring heads in different directions, which can be used for the quick label-free measurement of an object profile projected by structured light, and belongs to the technical field of optical measurement. In the calibration technology, only a calibration object of a plane with a plurality of calibration patterns is required to be designed and is not required to be manufactured into a high-precision three-dimensional target with known geometrical relationships on all surfaces; and geometrical parameters among all planes of the calibration object can be obtained by a close-range photogrammetry technology to upgrade the calibration object into the three-dimensional target, so the difficulty of machining and maintaining the high-precision three-dimensional target directly is reduced. By utilizing the technology, multiple structured light projected three-dimensional profile measuring heads can be measured from different visual angles in a label-free mode, and point clouds scanned in different directions are unified to one coordinate system automatically, so error accumulation caused by splicing is avoided effectively, and the time of the three-dimensional appearance reconstruction of the object can be shortened.

The invention discloses a tunnel construction informatization dynamic monitoring system and a monitoring method thereof. The system comprises an industrial personal computer, a PLC (programmable logic controller) wireless controller, a high-configuration digital camera, a deformation monitoring sensitization part, digital display acquisition equipment, wireless transmission equipment and a dynamic design feedback device. The method includes: 1), controlling commands to be transmitted to the PLC wireless controller by a computer control main station and controlling the digital camera and the digital display acquisition equipment; 2), acquiring internal force monitoring data by the aid of the digital display acquisition equipment; and acquiring a deformation monitoring image via the deformation monitoring sensitization part by the digital camera; 3), transmitting a monitored value to the computer control main station by the aid of the wireless transmission equipment; 4), analyzing results by the computer control main station via a real-time analysis module; and 5), transmitting the analyzed results to the dynamic design feedback device, analyzing relative difference of the really monitored value and a standard value, and early warming in real time by an early warning module. The tunnel construction informatization dynamic monitoring system is scientific, safe and practical, the method is convenient, instant, fast and effective, and safety problems in construction monitoring of tunnels are avoided.

Disclosed is a solid-state imaging device which includes a group of elements, the group including at least color photoelectric conversion elements configured to convert light signals in first, second, and third wavelength ranges to electric signals, respectively, a white photoelectric conversion element configured to convert light signals in the wavelength range including the entire visible light range and a portion of the infrared light range to electric signals, and a light-shielded diode element configured to be shielded from light. A unit is formed by including the white photoelectric conversion element and the light-shielded diode element for one color photoelectric conversion element, and within the unit, the white photoelectric conversion element is electrically connected with the light-shielded diode element by way of an overflow path. A camera provided with the solid-state imaging device is also disclosed.

The invention discloses an SOC estimation method based on a Kalman filtering method, and belongs to the field of power battery management. The method specifically comprises the steps that 1, a power battery model is built, and a precise lithium power battery model is built mainly by determining an equivalent circuit model and based on test data; 2, power battery SOC is estimated; 3, EKF algorithmis adopted to estimate an SOC result and conduct error analysis. The method provided by the invention analyzes the working principle of a lithium ionic battery and builds a power lithium ionic batterydivalent RC equivalent model, identifies the parameter models through laplace transformation and a large number of experiment data, and adopts a matlab platform to estimate the power lithium batterycharge state. The emulation result shows that the model has the good adaptability to the complex working conditions of electric automobiles, not only can the precision of SOC estimation be improved, but also the computing quantity is small, and SOC estimation is easy to achieve.

The invention provides a method for accurately reconstructing a microcosmic finite element grid model of a dissimilar material on the basis of CT (computed tomography) images. According to the method, sequence CT images are acquired through industrial CT, and micro-structural information in the CT images is mapped onto the reconstructed finite element grid model on basis of digitization and threshold segmentation, so that any detailed structural information in the dissimilar material can be represented in the reconstructed model. The method improves the reconstruction accuracy by means of contrast-limited adaptive histogram equalization, median filtering and pixel interpolation, and improves the reconstruction efficiency through image cut and pixel combination. With the method, rectangular (two-dimensional) and cuboid (three-dimensional) unit grid models with higher finite element analysis accuracy are directly reconstructed, error accumulation during reconstruction, grid partition and other links of the existing geometric reconstruction method is avoided, and reconstruction accuracy and efficiency are improved. The method can be widely applied to fields such as performance prediction and optimization design of dissimilar materials.

A supervisory control and data acquisition protocol converter having encoded meter register input ports which are able to interrogate, read, and retain data stored in viscous fluid or gas meter registers which communicate with the Automatic Meter Reading (AMR) protocols and convert that data upon request to an industrial monitoring and control system protocol such as MODBUS®, DF1, EtherNet / IP™, or ADAM 4000. The present art further provides serial RS-232C and RS-485, and Ethernet ports whereby said data may be transmitted and / or the apparatus may be setup according to user defined specifications. The present art apparatus also provides industrial standard pulse (0-5 volts) and current (4-20 milliamps) inputs. The present art further provides a user configurable display which may display any or only a portion of the aforesaid data along with other user desired data. The apparatus further serves a web page over the Ethernet port with the aforesaid data populating a table.

The invention relates to an automatic extraction method for newly-increased construction land image spots in high-resolution remote sensing images based on the NDVI and the PanTex index. The method includes the first step of inputting a front time phase high-resolution remote sensing image and a back time phase high-resolution remote sensing image, and then conducting geometric fine correction and relative radiation correction, the second step of calculating a front time phase NDVI image, a back time phase NDVI image, a front time phase PanTex image and a back time phase PanTex image, the third step of conducting unsupervised clustering on the two NDVI images and the two PanTex images respectively, the fourth step of extracting a binary change image from vegetation to buildings, bulldozed and filled earth according to the two NDVI clustered images, the fifth step of extracting a binary change image from vegetation, bulldozed and filled earth to the buildings according to the two PanTex clustered images, the sixth step of extracting interfering ground object regions, the seventh step of conducting union operation on the two extracted change images and removing interfering ground object masks to obtain a newly-increased construction land image, the eighth step of segmenting the back time phase images, and the ninth step of calculating the proportion of changed pixels in each segmented image spot and extracting the newly-increased construction land image spots. Through the method, three types of newly-increased construction land image spots can be effectively extracted, and auxiliary information can be provided for land utilization change investigation.

A particulate matter accumulation amount detection apparatus includes a filter pressure loss detecting section, an incombustible component amount estimating section, an incombustible component accumulation state estimating section, and a particulate accumulation amount detecting section. The filter pressure loss detecting section is configured to detect an upstream-downstream differential pressure of an exhaust gas purification filter having a plurality of substrate walls. The incombustible component amount estimating section is configured to estimate an amount of an incombustible component. The incombustible component accumulation state estimating section is configured to estimate an accumulation state of the incombustible component on the substrate walls based on the engine operating state and the amount of the incombustible component. The particulate accumulation amount detecting section is configured to detect an amount of the particulate matter accumulated in the exhaust gas purification filter based on the upstream-downstream differential pressure and the accumulation state of the incombustible component.

The invention discloses a polarized SAR (Synthetic Aperture Radar) image semi-supervised classification method capable of considering characteristic optimization. The method comprises the following steps: firstly, adopting a refined polarized LEE filtering method to carry out filtering, extracting polarization characteristics, carrying out combination to obtain an original characteristic set, and carrying out normalization processing; selecting an initial training sample set and a no-label set, and carrying out characteristic selection and classifier parameter optimization through a hybrid coding genetic algorithm under the initial training sample set; reconstructing the training sample set and a no-label sample set; training the classifier, and selecting a candidate set from the no-label sample set; utilizing a trained SVM (Support Vector Machine) classifier to label the candidate set, and selecting and expanding sample points with a high confidence coefficient into the training sample set; repeating the training of the classifier until learning is finished; and classifying the whole image by a finally trained SVM to obtain a classification thematic map. By use of the classification method, on one hand, effective characteristics can be adaptively extracted to improve a semi-supervised classification effect; and on the other hand, the efficiency of self-training learning can be improved, and error accumulation can be effectively avoided.

The embodiment of the invention discloses an image processing method, framework and device in the computer vision technology in the field of artificial intelligence, and the method comprises the steps: obtaining multiple frames of RAW images, carrying out image alignment, channel splitting and pixel recombination of the obtained multiple frames of RAW images, carrying out detail recovery of the images based on a deep learning network; and enhancing brightness and colors of images outputted by the deep learning network. According to the embodiment of the invention, various types of processing related to detail recovery are fused in the same deep learning network, thereby avoiding the mutual influence between different types of processing when the various types of processing are carried outin series, effectively fusing the effective information of a plurality of frames of images, and improving the image quality. Image processing effect can be better improved.

The invention relates to an on-wafer testing method for a micro-wave element based on an unknown material substrate. The method comprises the following steps: firstly, a TRL calibration part, an open circuit resonator and an element to be measured which are suitable for micro-wave on-wafer testing are prepared on a substrate; secondly, a micro-wave on-wafer testing system is calibrated by using a TRL calibration method; thirdly, resonation point of the open circuit resonator is tested by the calibrated micro-wave on-wafer testing system; fourthly, effective dielectric constant and transmission line impedance of the substrate are calculated according to a measured resonation point; fifthly, an S parameter of the element to be measured is measured, and reference impedance of the S parameter is the impedance obtained in the last step; and finally, a reference S parameter with 50 omega impedance is acquired by adopting a method of S parameter impedance conversion. The method has the advantages that the system impedance of the TRL calibration part can be calculated through testing the open circuit resonator; the method combined with standard TRL calibration not only can carry out on-wafer testing on the micro-wave element under the condition that the characteristic of a material is unknown, but also can be combined with a method of field analysis to calculate the real characteristic of the material.

The invention provides an oppositely-closing attitude-adjusting method based on a multi-laser tracker measuring field. The oppositely-closing attitude-adjusting method comprises the steps that S1, anideal assembly model is established; S2, the multi-laser tracker measuring field is constructed; S3, a bundle adjustment method is adopted for iterative computation, and a homogeneous transformation matrix between any two laser trackers is solved; S4, coordinates of all enhanced reference systems (ERS) under an overall coordination system are calculated; and S5, target workpiece tools are assembled onto standard workpiece tools. Position selection of the ERSs is small in constraint and more flexible, drifting of the relative positions between all the ERSs is avoided, and the measuring precision is improved. Station transferring computation adopts the bundle adjustment method based on an overall optimization idea, station transferring computation between the multiple laser trackers can be completed at a time, and the station transferring precision is improved. Meanwhile, actually measured data of the laser trackers are adopted to guide attitude adjusting, the influence of placing errorof target measuring points on the position estimation result of the target measuring points is reduced, and the measuring precision is improved.

The invention provides a microminiature inertial measuring unit, comprising an acceleration sensor unit, a signal conditioning circuit part, an A / D conversion module and a control and data processing part consisting of DSP chips, wherein the acceleration sensor unit comprises nine acceleration sensors arranged in a case and is positioned at the front end, output signals of the acceleration sensor unit are input to the A / D conversion module through the signal conditioning circuit; signals output by the A / D conversion module enter the control and data processing part consisting of DSP chips; and the control and data processing part comprises a data preprocessing module and an angular movement parameter calculation module; the data are preprocessed and relative movement parameters are then calculated and output by angular movement and linear movement parameter calculation methods. The device expands the application range of silicon micro-accelerometers, overcomes the shortcoming of low precision of silicon micro-gyroscope and calculates the angular velocity by using the measuring information of acceleration.

The invention discloses an intelligent automobile high-precision positioning and space-time situation safety method based on multi-sensor fusion. The method comprises a high-precision positioning method which comprises the steps of: firstly, preprocessing; secondly, initializing the front end of a visual inertia odometer; thirdly, optimizing the rear end of the visual inertia odometer; and fourthly, optimizing a global pose graph; and a space-time situation safety method. The method has the beneficial effect that the method can well adapt to a working scene and a developed fusion strategy. According to the invention, a key frame selection algorithm is realized, the algorithm can reasonably select different numbers of key frames according to different working conditions and can process theworking conditions where feature tracking loss is likely to occur, the space-time situation safety strategy fully considers the safety and traffic efficiency of roads, static obstacles and dynamic objects and the like, and the method has the advantages of high safety, high efficiency, small calculated amount, high real-time performance, low technical difficulty and the like.

The invention discloses a multi-starting-point incremental three-dimensional reconstruction method based on large-scale disordered images. The method comprises the steps of performing image matching, constructing a matching relationship chart, selecting a reconstruction starting point, detecting an edge, performing multi-starting-point three-dimensional reconstruction and splicing a submodel. By means of a clustering strategy and an iteration drift strategy, an image pair which is suitable as starting points is adaptively selected from a disordered image set and reconstruction is performed simultaneously; before reconstruction, determining an optimal reconstruction sub-process of each image according to a layered shortest path first algorithm, and determining a reconstruction boundary; and obtaining a whole three-dimensional reconstruction model according to sub-models obtained in different sub-processes and splicing the common parts contained in the sub-models. The three-dimensional reconstruction method according to the invention has advantages of settling a problem of suspension in reconstruction process when a current scene comprises a plurality of non-overlapped parts, obtaining a whole three-dimensional model which is covered by the image set, preventing three-dimensional structure error caused by matching and transmission of error images, realizing parallel processing in reconstruction sub-process, and improving image reconstruction efficiency.

The invention provides a triplication redundancy-based anti-radiation self-refreshing register. In two stages of latch structures of the register, a voter is added to the second-stage latch structure; an output end connected with the register is arranged at the input end of the voter; data of other two input ends are from other two outputs of the self-refreshing register with a triplication redundancy structure. During operation, when one of three data output paths is impacted by single particles to be turned, the rest two paths immediately correct a storage value for the register with a wrong storage value, so that the problem of turnover of two registers caused by radiation accumulation when the registers work in an irradiation environment for a long term is avoided.

The invention discloses an NRIET quantitive precipation estimation method based on cloud classification and machine learning. The quantitive precipation estimation method based on cloud classificationand machine learning comprises the steps of first preprocessing radar data and rain gauge data, and matching a radar reflectivity with rain gauge precipation data based on a site; identifying different cloud systems such as strati and convective clouds according to the radar reflectivity intensity; then performing fitting training in real time using a machine learning regression algorithm to obtain a relationship model between cumulative precipation and radar combined reflectivity; and finally, applying the relationship model between cumulative precipation and radar combined reflectivity to radar combined reflectivity lattice field data in real time to obtain a quasi-real-time quantitive precipation estimation field.

The invention discloses a micromechanical resonator and a manufacturing method thereof. The micromechanical resonator is formed by a resonator crystal plate and a packing cover piece through bonding and sealing. The resonator crystal plate comprises an input electrode, an output electrode, a bias electrode, a resonance unit and a supporting structure. Larger-area grounding holes around the resonator structure reduce feed-through signals, and micro heaters around the resonator achieve high-precision temperature compensation. The outermost peripheral structure of the resonator crystal plate is a bonding packing ring, and the packing cover piece comprises a packing cavity, the bonding packing ring and an electrical leading-out structure. According to the micromechanical resonator and the manufacturing method thereof, high-precision and large-scale manufacturing of the high-performance resonator is achieved, wafer-level air impermeability bonding packing is achieved, and the manufacturing method can be applied to low-cost large-scale production of high-performance MEMS resonators.

The invention belongs to the technical field of remote sensing image ground object change detection, and particularly relates to a remote sensing image ground object change detection method based on multi-task learning, which is used for solving the problem of error accumulation caused by the fact that the ground object change detection precision depends on the ground object classification precision in the prior art. A multi-task learning ground object change detection model is adopted and comprises two semantic segmentation model branches and a change detection model branch. A semantic segmentation model is constructed through a segmentation network, a feature extraction module of the model can effectively extract features of a remote sensing image, then a twin network is constructed to train a ground object change detection model, and a multi-task learning mechanism is constructed. In conclusion, the ground object change detection area can be determined, change detection results of different ground objects and ground object types before and after area change can be obtained, meanwhile, the problem of error accumulation is avoided, and the change detection precision is improved.

The invention discloses a grinding device of a crankshaft by one clamping, and belongs to the field of crankshaft grinding. Left and right supporting plates are vertically arranged on a rack, and left and right double rotating shaft positioning discs are respectively and coaxially arranged on the left and right supporting plates and are in a cylindrical shape; crankshaft clamping chucks and V-shaped block mounting grooves are arranged on the left and right double rotating shaft positioning discs, and the crankshaft clamping chucks and the double rotating shaft positioning discs are eccentrically arranged; the eccentric distance is equal to the radius of a crank of the crankshaft, and V-shaped blocks are arranged in the V-shaped block mounting grooves; an output shaft of a driving motor is coaxially connected with the right double rotating shaft positioning disc, and an output shaft of a movable indexing mechanism and the crankshaft clamping chuck on the left double rotating shaft positioning disc are coaxial. The invention also discloses a grinding method of the crankshaft by one clamping. The main journal and the rod journal of the crankshaft can be ground only by one clamping without a complicated numerical control system, and the operation is simple; the main journal and the rod journal of the crankshaft are theoretical cylinders, so that the principle error is avoided, and the machining accuracy is improved.

The invention discloses an indoor localization step size calculation method.The method includes the steps that an inertial measurement unit is arranged on the waist of a user, and inertial sensor data of the user is obtained in the indoor walking process of the user and includes accelerated speed and a quaternion; the quaternion is used for generating a coordinate-transformation matrix, and the accelerated speed based on an IMU carrier coordinate system is converted into a geographic coordinate system; triaxial resultant accelerated speed of gait detection is obtained according to triaxial accelerated speed in the geographic coordinate system; resultant accelerated speed data is processed to obtain a zero-speed correction point to complete gait detection, and the continuous motion of the user is divided into all single steps; an inverted pendulum model is used for calculating the distance of each single step of the user namely a single step size, the single step sizes are accumulated to obtain the total distance of user motion, and the motion step size is obtained.Error accumulation of quadratic integral is effectively avoided, and the accuracy of indoor localization step size calculation is effectively improved without affecting walking of the user.

The invention discloses an SLAM-based robot navigation method comprising a pose determination step, a sub-map construction step, a sub-map splicing step and a global map optimization step. According to the SLAM-based robot navigation method disclosed by the invention, sub maps are constructed and splicing is carried out to assist a robot in acquiring complete map information of an ambient environment. When the method is used for carrying out navigation for the robot, the high accuracy of the sensor data is ensured; and on the basis of multiple optimization strategies, the advantages of the laser SLAM navigation method can be realized to the greatest extent. In addition, the invention also discloses a robot using the method for navigation; and the own pose information and feature information of the ambient environment can be obtained, so that the high safety in the robot marching process is guaranteed.

The present invention relates to a seam line network generating method suitable for convex polygonal image effective ranges during mosaic of multiple digital orthophotos in an area range, comprising the following steps: obtaining an effective range of each orthophoto, and expressing the effective ranges with convex polygons; calculating segmentation lines between adjacent images; calculating an effective mosaic convex polygon of each image; and generating a seam line network. With the steps, mosaic treatment can be implemented on the basis of the generation of the seam line network; pixels contributing to mosaic in every image, seam lines relating to every image and images relating to each seam line can be conveniently obtained and a mosaic result can be directly generated; flexibility and efficiency of the mosaic can be guaranteed and error accumulation as well as intermediate results can be avoided, and a treatment result is independent of orders of the images. With the seam line network generating method, a problem of generating the seam line network for convex polygonal image effective ranges during mosaic of multiple digital orthophotos in an area range can be effectively solved.

The invention discloses a method, a device and equipment for detecting and recognizing a tire mold character and a storage medium. The method comprises: inputting a sample of a collected tire mold picture data set into an improved SSD detection network for training, and establishing a character detection model; using a convolutional neural network to extract a feature vector of a character regionimage detected by a character detection model and inputting the feature vector into a long and short time memory network for training to establish a character recognition model; performing target detection on a character detection region of a to-be-detected tire mold image by the character detection model; and outputting a recognition result of a large character or a small character string in thecharacter region by the character recognition model. In the character detection stage, the detection model is established for the characteristics of the tire mold, so as to achieve high detection precision on the mold image. In the character recognition stage, the features are fused in the long and short time memory network after being extracted, and the recognition model is established to realizenon-split and quick recognition of a large character and a small string of an indefinite character length.

The invention provides a Gaussian radial basis function based global sensitivity analysis method. A radial basis function having higher nonlinear function approximation accuracy serves as an approximation model, an analytic calculation formula of an Sobol index of the model is further obtained, and the sensitivity analysis accuracy for a complicated model can be effectively improved.

The invention discloses a large-power turning and milling composite machine tool, belongs to the technical field of machine tools, and aims to solve a problem in a traditional composite machine tool that a milling device is only used for auxiliary machining. The key point of the technical scheme is that the large-power turning and milling composite machine tool comprises a lathe bed, a main shaftinstalled on the lathe bed, a back face knife table installed at one side of the main shaft, and the milling device installed at the other side of the main shaft. The milling device comprises a driving mechanism and a milling mechanism rotatably installed on the driving mechanism. The large-power turning and milling composite machine tool is capable of, through installing the milling device with aB-axis on the lathe bed, and enabling the milling device to be rotated in multiple degrees, completing the milling and machining of a complicated workpiece by cooperating with the main shaft, effectively shortening a product manufacturing process chain, improving production efficiency, reducing a clamping frequency of the workpiece, and effectively improving machining precision, and further has the advantages of reducing floor area, and reducing production cost.

The invention relates to a method for establishing a vision / inertia combined navigation model based on a time delay filter. According to the method, when inertia navigation parameters are corrected byusing position and speed information of vision navigation, time delay errors of combined navigation are put into a filtering state equation, and through filtering estimation, not only are inertia navigation parameter errors estimated, but also time synchronization errors of the combined navigation can be obtained. The problem of out-synchronous combined navigation parameter resolution can be solved. The method has functions of inertia navigation resolution, vision navigation parameter resolution, time delay filter establishment, and the like. Compared with the prior art, the method has the advantages of being rigorous in theory, novel in method and efficient in performance.

The invention discloses a license plate recognition method based on deep learning, which comprises the following steps: step 1, training a deep convolutional neural network model M for license plate character detection; constructing a license plate character label set B; step 2, inputting the positioned license plate image I to a license plate character detection network M, and outputting a candidate license plate character set H; step 3, reordering the candidate license plate character set H obtained in the step 2 according to the abscissa hi.x of the upper left corner of the smallest circumscribed rectangle of the candidate license plate character from small to large to obtain a set C; step 4, further executing screening operation on the set C obtained in the step 3; step 6, sequentiallytraversing the set E obtained in the step 5; step 7, returning that license plate recognition result L obtained in the step 6. The method has the beneficial effects of effectively inhibiting the influence of license plate character adhesion, fracture, deformation, license plate dirt, license plate inclination, residual shadow on the license plate and the like.