209results about How to "Quick culling" patented technology

The invention discloses a ship multi-target detection method based on rotation area extraction, and the method comprises the steps of collecting a remote sensing image containing a ship target, and cutting the remote sensing image into a set size; carrying out rotation rectangle marking on ship the targets contained in the remote sensing images respectively, and taking the remote sensing images marked with rotation rectangles as a training set and a test set; inputting the training set into a ship multi-target detection model for training; inputting the remote sensing image to be detected intothe trained ship multi-target detection model for detection to obtain a preliminary detection result comprising a ship target pre-selection frame and the corresponding confidence; and calculating a rotation intersection-union ratio of the preselected frame with the highest confidence coefficient and other preselected frames, and deleting the preselected frame with the rotation intersection-unionratio greater than a set threshold to obtain a final detection result. By predicting the minimum area circumscribed rectangle of the ship target, the position, the length, the width and the axial information of the ship target are accurately outputted, the multi-target detection is realized, and the detection performance is remarkably improved.

The invention discloses an online thickness monitor and an online thickness monitoring method. The online thickness monitor comprises a CCD industrial camera, laser, a microprocessor and an interface cable, wherein the CCD industrial camera and the laser are connected with the microprocessor through the interface cable and carry out information interaction with the microprocessor through the interface cable; the laser and the CCD industrial camera are installed on a reference plane at a certain inclined angle, the laser transmits a laser beam, the CCD industrial camera captures a reflected beam, the laser beam transmitted by the laser is reflected by the reference plane, and the reflected beam falls at the center of a view field of the CCD industrial camera; and the reference plane is a horizontal table used for placing a measured object. The online thickness monitor disclosed by the invention can carry out high-precision monitoring and measurement on the thickness of products, thereby being convenient for operators to master the thickness of the measured object visually and conveniently.

The invention relates to an online grading method for tobacco based on a deep learning algorithm, which is characterized in that positive and negative classification, three-classification of positive group tobacco, green tobacco and sundry tobacco and then grading of the positive group tobacco are performed through a complete set of tobacco grading system, and the method is applicable to online grading; background masking is performed based on a method of supervised learning, and the background removing accuracy is higher compared with a traditional algorithm; classification and grading are performed by using all positive tobacco, and tobacco grading characteristics are more reflected on the positive tobacco, so that classification and grading performed by using the all positive tobacco have more rigorous scientificity and higher accuracy; grading for the positive group tobacco is performed by adopting a GoogLeNet model, and higher grading accuracy can be acquired by using more convolutions and larger number of network layers. The online grading method can perfectly achieves quick, lossless and online rejection of the green tobacco and the sundry tobacco, accurately divide the grade of the positive group tobacco and ensure all aspects of benefits of purchasing to be not damaged.

It is an integration circuit chip vision aligning method belonging to integration circuit chip seal method, which comprises the following steps: to select sample from the lead wire frame and chip; to get the sample measurement coordinates by image process and to measure the defect chip by image constant moment; to get the chip position transformation parameters to get the coordinates; to get the sample non-linear error by minus measurement coordinates from computation coordinates till the error is within the permitted value; finally to transform the bonding arranged coordinates into bonding workbench displacement coordinates.

The invention provides a GM-HMM (Gaussian Mixture-Hidden Markov Model) driving behavior prediction method based on visual characteristics. A six-degree-of-freedom driving simulator and an eye tracker system are selected to carry out a simulation experiment, a fuzzy K-means dynamic clustering algorithm is adopted to divide the distribution of a driver interest visual field area, and the Pauta criterion of an improved Bessel formula is adopted again to remove abnormal line of sight points without a definite boundary among divided driver interest visual field areas; then, from an entry point of the visual representation parameter of a driver, a container type graph analysis method in mathematical statistics is adopted to verify the difference of the parameter through a NEMENYI rank sum test in SPSS (Statistic Package for Social Science) software, and the visual representation parameter sequence of the driver is determined through R type index clustering; and finally, a GM-HMM driving behavior prediction model is established, and the reliability of the GM-HMM driving behavior prediction method is analyzed.

The invention provides a data processing method, device and equipment of game scene, and a readable storage medium. The method comprises the following steps: performing map combination of multiple to-be-processed models in the game scene, enabling the textures and the maps of various models to be consistent, thereby facilitating the grid combination of the models; performing voxelization preprocessing on multiple models by computing bounding boxes of multiple models to obtain the voxelization data of multiple models; determining redundancy panels in multiple models according to the voxelization data of multiple models; rejecting the redundancy panels in multiple models, thereby offline and quickly identifying and rejecting the redundancy panels in multiple models in the whole scene, and reducing the DP and Overdraw produced in the rendering process; the complexity is low, and the rendering efficiency and performance in the running can be improved.

The invention discloses a satellite-borne image processor for remote sensing image target characteristic recognition. The satellite-borne image processor comprises a first image processing unit, an interface module and a power conversion module, the first image processing unit comprises an operation module and a storage module; the storage module is used for storing real-time remote sensing imagedata and base map data of the to-be-detected area; the operation module comprises an FPGA module and a DSP module which are electrically connected with the storage module. The FPGA module is communicated with the DSP module through an EMIF bus. The FPGA module is used for receiving and outputting real-time remote sensing image data of a to-be-detected area; and the DSP module is used for carryingout target detection, feature extraction and positioning calculation on the remote sensing image data output by the FPGA module. The method has the characteristics of low power consumption and high detection efficiency, and can adapt to severe satellite environments.

The invention relates to the field of digital image processing and provides a random sample consensus (RANSAC) algorithm for removing mismatching pairs from image splicing after characteristic point matrices of images are compressed. The algorithm can increase operational speed greatly, is suitable for parallel computation, can greatly improve the matching efficiency, and provides an optional image splicing and fusion solution for diverse multimedia equipments at present. Therefore, the invention adopts a technical scheme that an image matching method utilizing block Boolean operation comprises the following steps of dividing the matrices, compressing the matrices and conducting Boolean conversion, matching characteristic matrix blocks in Boolean formats of the two images, searching the best characteristic matching pair, conducting matching operation of the RANSAC algorithm, and finally confirming the best affine matrix. The image matching method is mainly applied to digital image processing.

The invention relates to an LED chip maintenance method. The LED chip maintenance method comprises steps that S1, a chip substrate having a failed LED chip is provided; S2, the position of the failedLED chip on the chip substrate is determined; S3, a first nozzle is heated to a preset temperature, and the first solder for fixing the failed LED chip is melted; S4, the failed LED chip is sucked; S5, the first solder is sucked; and S6, die bonding is performed. The method is advantaged in that when the LED chip fails, the first nozzle is heated to fuse the first solder connecting the failed LEDchip with the chip substrate, after the first solder is melted, the failed chip is sucked, the first solder is removed by suction, so elimination of a failure point of the chip substrate is simple, fast, and efficient, professional equipment is not needed, cost is low, moreover, the failed chip and the first solder can be removed, the quality of subsequent die bonding is guaranteed, and the methodis especially suitable for maintenance of small-pitch LED products such as COB flip chips.

The invention discloses a self-measurement-control mosquito expelling and killing device which comprises a mosquito sound positioning assembly, an ultrasonic expelling assembly, a mosquito sucking and killing assembly and an assembly control module. The assembly control module is connected with the mosquito sound positioning assembly, the ultrasonic expelling assembly and the mosquito sucking and killing assembly. The mosquito sound positioning assembly comprises audio sensors, an operational amplifier module and an AD module which are sequentially connected. The assembly control module calculates the sound source position and the frequency and phase of a sound source through the different linear distances between at least two audio sensors in the mosquito sound positioning assembly and the sound source and drives the ultrasonic expelling assembly or the mosquito sucking and killing assembly to work according to the distance, frequency and phase information. When the ultrasonic expelling assembly works, the ultrasonic expelling assembly gives out same-phase or reverse-phase low-frequency modulation ultrasound wave signals or low-frequency signals with the same frequency as that of the sound source of mosquitoes. The self-measurement-control mosquito expelling and killing device can react on mosquitoes of different types and genders in different regions and reach the mosquitoes intelligently to achieve the optimal culling effect; the mosquitoes can be expelled or killed intelligently, efficiently and quickly through sound-light alarm.

The invention discloses a method for testing a mono-pulse radar horizontal direction diagram based on the pulse sorting technology. The method mainly comprises the following steps that S1, according to a set sampling frequency, signals of radar to be tested are sampled so that a sampling point sequence can be obtained; S2, a pulse threshold is calculated; S3, a pulse sequence is determined and the pulse time, the pulse width and the pulse amplitude of each pulse are calculated; S4, pulse width sorting is conducted on the pulse sequence; S5, pulse time interval sorting is conducted on the pulse sequence; S6, a rotating period of the radar is calculated; S7, point supplementation is conducted on the pulse sequence; S8, the radar horizontal direction diagram is formed. The method has the advantages that the radar horizontal direction diagraph can be tested under the condition that the radar does not stop working, sampling is conducted on the received signals at the high sampling frequency, so that the accuracy of an antenna direction diagram is improved, and fault diagnosis and maintenance of the mono-pulse radar are facilitated.

The invention relates to a method used for rapid evaluation of priming paint performance. The method comprises following steps: the surface of an asbestos cement plate and the surface of a fiber reinforced cement middle density flat board are coated with a priming paint to be tested respectively; the asbestos cement plate and the fiber reinforced cement middle density flat board are subjected to water permeability test, adhesion test, alkali burnout resistance test, and alkali burnout resistance and salting-out resistance test; if the water permeability, adhesive force, the alkali burnout resistance of inner wall priming paint, or alkali burnout resistance and salting-out resistance of outer priming paint are all qualified, the priming paint is qualified; if one parameter selected from the water permeability, adhesive force, the alkali burnout resistance of inner wall priming paint, or alkali burnout resistance and salting-out resistance of outer priming paint is not qualified, the priming paint is unqualified.

The invention discloses an automatic and continuous filling system of blood sampling test tubes and relates to the technical field of test tube filling equipment. The problems that during filling, test tubes are easily abraded, and defective products of the test tubes cannot be automatically detected are solved. A machine body, vibration disks, guide pipe channels, a roller bed, a collecting box,a vibration tray, a mold frame, a tube installing mechanism, a transferring mechanism and a conveying mechanism are included; and a linear array camera, pneumatic fingers and a vacuum suction nozzle are arranged at the position, over the roller bed, on the machine body, and a push rod is fixed to the pneumatic fingers. According to the automatic and continuous filling system of the blood samplingtest tubes, the test tubes are automatically laid flatly through the roller bed, the conveyed test tubes are subjected to line scanning image comparison through the linear array camera, defective products of the test tubes are positioned, and are pushed out under the joint effects of the pneumatic fingers and the vacuum suction nozzle, and the quality of the filled test tubes is guaranteed; and due to the manner of roller bed guide pipes, the test tubes are arranged smoothly and orderly, the tube clamping phenomenon is reduced, and damage to the test tubes is further reduced.

The invention discloses a battery screening method and device, a computer storage medium and a test system. The method comprises that a battery to be tested is charged in constant current according tothe present current, and a corresponding practical temperature change value in the formation process of the battery to be tested is determined; and the practical temperature change value is comparedwith a preset temperature range to determine whether the battery to be tested is qualified. The preset temperature range is determined on the basis of temperature change values corresponding to samplebatteries with the same nominal capacity in the formation process.

The invention provides a loopback detection method and system, a readable storage medium and electronic equipment, and the method comprises the steps: receiving node data, carrying out the Hash codingbased on a GPS position in the node data to obtain a key value, and achieving the nearly real-time loopback retrieval efficiency while enabling the global position (GPS) information received by a vehicle to be utilized separately; searching a historical node set which has the same key value and comprises at least one historical node from a historical GPS position code library, screening a targetloop node from the historical node set, constructing a loop based on the target node and the target loop node, and rapidly eliminating a pseudo loop formed by long-time parking by adopting the constraint of time and track length, thereby improving the precision of track optimization.

The invention relates to a dispatching optimization method based on a two-layer nest structure and belongs to the technical field of industrial automation. A nest optimization algorithm structure is adopted for the problem of mixed integral non-linear programming (MINLP) existing in the production dispatching of most of manufacturing enterprises, global search is conducted by means of continuous updating of multiple solutions without the gradient information of a solution space, a feasible coarse judgment model is added and constructed, the feasibility of a solution is coarsely and fast judged before the solution evaluation, so that it is ensured that the true satisfactory solution of the dispatching model is obtained with a high probability, the calculated amount is small, the optimizing process is simple, and the application range is wide.

The invention relates to a semantic SLAM method based on GMS feature matching in a dynamic scene, and the method comprises the steps of carrying out the segmentation of an input image through a semantic segmentation network, obtaining the masks of all objects, removing the mask part of a dynamic object, and obtaining a preliminary image after the dynamic object is removed; extracting ORB feature points from the input image, and then calculating a descriptor; detecting and removing dynamic feature points according to a method of combining motion consistency and semantic information; through themode of removing the dynamic state by combining the motion consistency and the semantic information, improving the operation precision and robustness of the visual SLAM system in a high-dynamic environment.

The invention discloses a head detection method and a device, which is used for detecting head images from input original images, as well as a head detection and category judgment method and a device, which is used for detecting head images from input original images and carrying out category judgment. By using the embodiment of the invention, region of interest (ROI) images which are not the head images obviously are quickly removed from the original images by a first-level head detection based on TGF characteristics and then head images are selected from the rest ROI images by a second-level head detection based on shape characteristics. In addition, if required, confirmation and category judgment can be further carried out on the ROI images which are head images based on results of two levels of head detections to distinguish front / profile / back head images or even non-head images which are detected as head images by mistake. The invention can secure low omission rate, improve accuracy of head detection and reduce false detection rate.

The invention discloses an adhesive tape foreign matter eliminating device which comprises a supporting mechanism and an eliminating mechanism, wherein the supporting mechanism comprises a base and a first bracket and two second brackets which are separately arranged on the base; the two second brackets are used for supporting a conveyor belt and are separately arranged on two sides of the first bracket; a sliding assembly, through which the first bracket moves vertical to the moving direction of an adhesive tape with a foreign matter, is arranged on the base; the eliminating mechanism comprises a lower module fixed at the bottom of the first bracket, an upper module located right above the lower module, and an air cylinder fixed at the top of the first bracket; and the output end of the air cylinder is connected to the upper module to drive the upper module to move up and down relative to the lower module. The adhesive tape foreign matter eliminating device provided by the invention can quickly eliminate metal foreign matter portions contained by the adhesive tape completely, so that the detection speed and the tire production efficiency are increased.

The invention discloses an indoor target detection method and device. Depth image information and color image information of an indoor target are acquired; the depth image information and the color image information are fused by using the weight of the depth image information and the color image information so that the fused image information is obtained; and the fused image information is detected so that the detection result is outputted. According to the indoor target detection method and device, the color image and the depth image are combined, and the weights of the color image and the depth image are self-adaptively adjusted to be fused so that respective defects can be effectively compensated and the detection effect can be greatly enhanced. The indoor target detection method and device can be applied to embedded equipment real-time figure detection.

The invention discloses a method capable of quickly removing dynamic obstacle artifacts from a static background map in real time, and the method comprises the steps: collecting laser radar, IMU and GPS data, and converting n frames of radar point clouds into a UTM coordinate system, obtaining the kth frame of dense point cloud map by registering N-1 frames of point clouds before the kth frame tothe Nth frame, and for the frames of k being larger than 2N-2 frames, obtaining the kth frame of dense point cloud map by registering N-1 frames of static point clouds before the kth frame to the Nthframe; performing ground segmentation on the kth frame of dense point cloud map and the kth cloud, and reserving non-ground points No and Ni; rasterizing the point clouds No and Ni, and judging the dynamic and static attributes of each grid; and registering the static point clouds obtained by processing from the Nth frame to the (n-1) th frame to the nth frame of point clouds to generate a point cloud map which is the multi-frame point cloud dense map for eliminating the dynamic obstacle artifacts. The method greatly reduces the calculation amount while ensuring the precision, saves the calculation time, and achieves the purpose of quickly generating the dense point cloud map for eliminating the dynamic obstacle artifacts.

The invention relates to a rapid propagation method of cassava virus-free seedlings. The method comprises the following steps: step one, collecting explants and carrying out a detoxification pretreatment; step two, disinfecting the surfaces of explants and taking off the stem tips; step three, inducing the stem tips to obtain seedlings; step four, subjecting the seedlings to subculture multiplication culture; step five, carrying out virus detection through a semi-quantitative RT-PCR method; step six, domesticating and transplanting the virus-free tube seedlings. The provided method can effectively remove various viruses and bacteria taking cassava stems as the nutrient source, and obtains virus-free cassava seedlings. The healthy planting, purification and rejuvenation of cassava seedling can be achieved. The occupied space for propagation is small, the propagation speed is quick, and the survival rate of transplanting is high. The provided method can be applied to safe species exchange among different countries and areas, can provide healthy cassava seedlings, reduces the happening and spreading of diseases, and provides technical guarantees for sustainable development of cassava.

The invention relates to a bottle inspection and elimination device which can rapidly and stably eliminate unqualified bottles and does not influence the passing of qualified bottles. The bottle inspection and elimination device comprises a plurality of elimination fingers which are sequentially and adjacently arrayed along the conveying direction of bottles from short to long, wherein when bottles are detected to have no cover, have inclined covers or have no aluminum film, the plurality of elimination fingers are forwards turned to be in the horizontal state from the vertical state in sequence and then are reset; the longest elimination finger is used for conveying the bottles into an entrance of a rotary table type recycling mechanism so as to stably and rapidly eliminate the unqualified bottles and prevent the bottles from falling down; meanwhile, the subsequent passing of qualified bottles is not influenced.

The invention discloses a space debris detection method based on stellar retrieval. The method comprises: receiving images and estimating image backgrounds; performing full-frame scanning on the images according to the estimation results of the image backgrounds, and calculating two-dimensional plane coordinates of all the stars on the images; acquiring three-dimensional vectors of inertial spacecorresponding to all stars by combining the actually measured coordinates of the two-dimensional planes of all stars on the images by means of astronomical positioning and acquiring two-dimensional tangent plane theoretical coordinates of all fixed stars; identifying and rejecting background fixed stars in all images; and acquiring track features of the remaining stars, and detecting all space debris in the images. Therefore, the fixed stars in the single-frame images can be identified and rejected based on preset fixed star identification conditions; multi-frame images can be used flexibly; and the space debris of the image is detected, so that the detection false alarm caused by factors like the bright fixed star is reduced; and the detection success rate of the weak space debris is improved.

The invention relates to a BE soft box hard carton cigarette packet defect detection structure in a BE soft box hard carton packaging unit, in particular to a small box cigarette packet visual detecting and removing device which is used for being connected between a soft box packaging machine and a small box transparent paper packaging machine. The small box cigarette packet visual detecting and removing device comprises a cigarette packet detecting and conveying system arranged on the rack, an appearance visual detection system arranged on the rack and located on the periphery of the cigarette packet detecting and conveying system and a defective cigarette packet removing system arranged on the rack and located at the tail of the cigarette packet detecting and conveying system. The appearance visual detection system is connected to the defective cigarette packet removing system. The cigarette packet detecting and conveying system comprises an upper planar conveyor belt and a lower planar conveyor belt which rotate at the same speed. The conveying speed of the upper planar conveyor belt and the lower planar conveyor belt is higher than the cigarette packet conveying speed of a small box cigarette packet packaging machine. The distance between every two front-back adjacent cigarette packets is increased through speed difference, cigarette packets are prevented from being blocked mutually and being affected mutually. The effects that the shapes of cigarette packets are fixed, and the cigarette packets are conveyed stably are achieved. Accordingly, the stability and high efficiency are improved. The small box cigarette packet visual detecting and removing device has high accuracy.

The invention discloses a tensile test method and device for lithium battery cap solder joints. The device includes a platform mechanism, a tension mechanism arranged on the platform mechanism, and an unqualified cap discharge mechanism arranged on the platform mechanism; the platform mechanism includes a main frame , the transverse guide rail assembly arranged on the upper end of the main frame, the sub-frame arranged on the transverse guide rail assembly, the transmission assembly connected to one side of the sub-frame, the power part connected to the transmission assembly and the electric cylinder arranged on the upper end of the sub-frame , the power components are set on the main frame through the fixed plate. The above-mentioned device and method for testing the tensile force of the solder joints after cap welding in the production process of the lithium battery, through the monitoring of the tensile testing device and the comparison and calculation of the tension adjustment value, as well as the rapid movement and stroke setting of the electric cylinder, comprehensively, accurately and quickly determine the cap and the lithium battery. Excellent welding quality; through the quick response of the power unit and the linkage of the eccentric mechanism driven by the movement, the unqualified products can be quickly eliminated, which improves the quality and efficiency of monitoring.

The invention discloses a centering module and a full-automatic brick pasting machine thereof. The full-automatic brick pasting machine comprises the centering module, a rack module, a feeding moduleand a tile pasting module, wherein the centering module is used for centering ceramic tiles needing feeding so as to facilitate subsequent conveying and detection of the ceramic tiles; the rack moduleis used for outputting the ceramic tiles from the centering module and carrying out photographing and image recognition on the ceramic tiles through an industrial camera, the broken ceramic tiles areremoved through the removing mechanism, the ceramic tiles are finally conveyed above a feeding push plate of the feeding module, the rack module is further provided with a mortar mechanism, and the mortar mechanism is used for leveling mortar; the feeding module is used for pushing the ceramic tiles sent by the rack module into a feeding frame through the feeding push plate, and the ceramic tilesare gradually stacked from bottom to top in the feeding frame; and the tile pasting module is used for sucking the ceramic tiles in the feeding frame and then installing the ceramic tiles on a presetposition to finish tile pasting. According to the centering module and the brick pasting machine, automatic centering, conveying, feeding, sucking and tile pasting of the ceramic tiles can be achieved, the mortar can be laid automatically, and the whole tile pasting process can be automatically completed.

The utility model discloses a label printing quality detecting device. The label printing quality detecting device comprises a machine frame, and a label printer, a conveying device, a visual system, a marking device and a controller which are installed on the machine frame, wherein the conveying device is used for conveying a label paper tape forwards by pulling, and a detection zone is arranged on the conveying device; the visual system is used for shooting a bar code, characters and all images on a current label and sending an image signal to the controller; the controller is used for operating the image signal, judging whether the current label is qualified and sending a corresponding action command to the marking device; the marking device is positioned at the downstream of the detection zone and is used for marking an unqualified label according to the action command of the controller. The label printing quality detecting device detects a printed label through the visual system, and marks the unqualified label, so that an aim of quickly and accurately rejecting the unqualified label is achieved; compared with manual rejection, the automatic label printing quality detecting device has the advantages of higher efficiency and high accuracy and can meet the need of large-scale production.

The invention relates to a thrust distribution optimization method considering propeller performance. The method comprises the following steps of (1) forming a ship stress model in a matrix form, (2)receiving an expected thrust signal, (3) carrying out noise judgment, (4) solving the rotating speed of each propeller and the rudder angle of each rudder by using a single-step planning method, (5) analyzing whether a reasonable rotating speed and a reasonable rudder angle can be generated or not, if yes, entering the step (7), and if not, entering the step (6), (6) performing multi-step planning, dividing the expected thrust signal into a plurality of steps, and respectively solving the rotating speed of each propeller and the rudder angle of each rudder by using a single-step planning method, and (7) transmitting the solved rotating speed of each propeller and the rudder angle of each rudder to the propeller and a controller of the rudder. Unreasonable expected signals can be eliminated, generated thrust signals can adapt to the performance of the thruster, the rationality of a distribution result is improved, and the dislocation phenomenon of ship dynamic positioning is reduced.

The invention relates to a method for high-efficiency potato cross breeding with florescence prolonging and fructification promoting, which belongs to the technical field of potato breeding. The method solves the technical problems of short potato florescence, less blooming, and low fructification percentage. The method comprises the following steps: 1) arranging an overhead net house, and arranging sprinkling equipment; 2) accelerating germination of the potato; 3) sowing the potato and making frames; 4) performing artificial cutting on stem tubers; 5) collecting pollen and performing emasculation; 6) performing pollination and collecting fruits; and 7) taking seeds and performing storage. The method provides more convenient, faster and effective cross breeding work for the potato, the florescence of the early mature potato can achieve 25-30 days, the florescence of the later mature potato is 30-65 days, the flowering amount is greatly increased, the labor injury is reduced, and the work efficiency is greatly increased. The method is used for high-efficiency potato cross breeding with florescence prolonging and fructification promoting.