73 results about "Laser triangulation" patented technology

The application relates to the technical field of welding automation and robot visual perception, and discloses a robot intelligent welding method and system, wherein the method comprises the following steps: constructing a three-dimensional point cloud image by adopting a laser triangulation method; filtering and enhancing the point cloud image, and extracting a weld point cloud set by adopting a U-Net semantic segmentation model; identifying a space path and regressing a pose based on the set to obtain a position-pose pair trajectory primitive; fitting the path by adopting a cubic B-spline curve and generating a continuous welding trajectory by interpolation; and combining visual feedback in the welding process, introducing proportional differential control logic to compensate for closed-loop trajectory errors. Compared with the technical problem that the existing technology cannot stably realize weld path identification and accurate tracking under complex structures such as curved surfaces and variable cross sections, the application realizes high-precision automatic welding of complex weld structures by constructing a perception control closed loop based on visual servoing, and improves trajectory fitting precision.

This utility model discloses a laser triangulation ranging device and system; relating to the field of laser ranging technology; including a laser for emitting laser light towards the object being measured; a light-receiving lens assembly for receiving the reflected laser light from the object being measured and correcting and focusing the reflected laser light spot; the light-receiving lens assembly includes a focusing lens and multiple correction elements; the light spot of the reflected laser light from each correction element is corrected in different properties and falls onto the light-receiving element; this solution improves the structure of the traditional laser triangulation ranging optical path by correcting the light spot of the reflected laser light spot in different properties through the light-receiving lens assembly, ensuring measurement accuracy, reducing costs, and effectively accelerating assembly efficiency; by increasing the light spot size in the laser transmission direction at the position of the light-receiving element through the light-receiving lens assembly, the center of the light spot does not need to be aligned with the center of the light-receiving element to be detected, reducing the structural accuracy of the laser triangulation ranging device and improving assembly efficiency.

The utility model discloses a fin thickness online detection device, belongs to the technical field of fins, solves the problem of inconvenience in heat dissipation of a sodium ion battery, and comprises a base, and the upper surface of the base is fixedly connected with a moving mechanism, a cleaning mechanism and a laser triangulation measurer. The moving mechanism comprises a supporting rod fixedly connected to the upper surface of the base. A water pump can be started, cleaning liquid in a cleaning water tank can be pumped into a fixed pipeline through the water pump, then the cleaning liquid is sprayed out from a spray head, oil dirt impurities on the outer surface of a fin can be dissolved conveniently, meanwhile, a second servo motor is started, and an output shaft of the second servo motor drives an eccentric wheel to rotate; the eccentric wheel rotates to extrude the outer surface of the moving frame, so that the moving frame drives the two cleaning brushes to reciprocate, the outer surface of the fin is conveniently rubbed and cleaned, then the fin passes through the lower part of the laser triangulation measurer, and the accuracy of fin thickness detection is conveniently improved.

The invention discloses a correction device for carbon fiber badminton racket processing, which comprises a correction platform and a racket body, a correction structure and a detection structure are arranged on the top surface of the correction platform, a base is fixed on the bottom surface of the correction platform, a lifting piece is arranged on the front surface of the base, and rackets with different sizes can be adaptively clamped through a V-shaped racket frame and roller structures on the two sides; and the rotating shaft is kept stable during rotation. In combination with a laser detection end, a laser triangulation ranging principle is utilized to perform omnibearing scanning on a rotating racket body, and the physical form deviation of a racket rod and a racket frame is converted into a high-precision digital signal in real time, so that the bending point and the bending amount are automatically and accurately positioned in a non-contact manner, the traditional manual visual inspection is effectively replaced, and the detection efficiency and the reliability are remarkably improved.

This utility model discloses a brake disc testing device for new energy vehicles, including a platform. A control panel and a three-jaw chuck are respectively installed on the top of the platform. A transverse adjustment rail is rotatably mounted above the platform around the central axis of the three-jaw chuck. Laser triangulation sensors are fixed at equal intervals on the transverse adjustment rail via a fixing mechanism. A distance-fixing mechanism is provided on the transverse adjustment rail, equidistant from the bottom laser emission end of the laser triangulation sensor to the top of brake discs of different thicknesses. In this brake disc testing device for new energy vehicles, rotating a second rotating shaft causes the distance-fixing rod and several equally spaced laser triangulation sensors to move downwards simultaneously. When the lower end of the distance-fixing rod contacts the top of the brake disc, the distance from the top of the brake disc of different thicknesses to the bottom laser emission end of the laser triangulation sensor is equal, ensuring the accuracy of brake disc flatness detection.

The application discloses a kind of smooth ring gauge automatic calibration device and method, the device includes clamping positioning unit, visual identification unit, measuring unit and computer control system, measuring unit includes metrology coordinate measuring machine, automatic calibration module and temperature acquisition unit, visual identification unit includes line laser vision sensor and image processing module, computer control system is used to realize the data transmission and cooperation between each module or unit, line laser vision sensor is scanned to point cloud data using laser triangulation to smooth ring gauge 3D profile, image processing module identifies and analyzes point cloud data, obtains the geometric characteristic information of smooth ring gauge, automatic calibration module adjusts measurement scheme in real time according to geometric characteristic information, metrology coordinate measuring machine executes the full parameter measurement of smooth ring gauge according to adjusted measurement scheme.The application can realize the full parameter, high efficiency, high precision, integrated automatic calibration of smooth ring gauge.

The application provides a multi-category plastic automatic reflow sorting method and system and a storage medium. The method comprises the following steps: obtaining point cloud data of a bottle body, processing the point cloud data by using a laser triangulation method to obtain bottle body shape characteristics of the bottle body; obtaining initial spectral information of the bottle body and working temperature and humidity of a working environment, determining whether compensation of the spectrum is needed based on the working temperature and humidity and the initial spectral information, if needed, substituting the working temperature and humidity and / or the initial spectral information into a preset compensation model to obtain spectral compensation information, and determining actual spectral information of the bottle body based on the initial spectral information and the spectral compensation information; substituting the bottle body shape characteristics and the actual spectral information into a hybrid neural network model to obtain a plastic type of the bottle body and a corresponding confidence degree, determining a confidence level corresponding to the confidence degree, and generating a corresponding sorting instruction based on the confidence level to realize sorting of the bottle body. The application can efficiently and accurately realize multi-category plastic sorting work.

The invention discloses an automatic detection device for surface defects of synthetic leather, and belongs to the technical field of synthetic leather production lines. The front cleaning mechanism is arranged on the synthetic leather body for cleaning before the detection process, so that defects are exposed; the defect detection and recognition assembly is erected above the conveying mechanism and comprises a linear array camera directly facing the to-be-detected synthetic leather body, a thermal infrared imager arranged side by side with the linear array camera and a laser triangulation 3D sensor located on one side of the linear array camera and the thermal infrared imager and capable of adjusting the angle according to measurement requirements; and the flattening mechanism is arranged above the synthetic leather body, and the flattening mechanism can flatten elastic wrinkles and curls generated due to self elasticity or internal stress release of the synthetic leather body so as to assist identification and detection of the defect detection and identification assembly. Through the design of the front cleaning mechanism, the surface of the synthetic leather is cleaned more thoroughly, and the detection accuracy is improved.

The application discloses a laser ranging four-roller micro-foaming calender and a calendering method thereof, and belongs to the technical field of plastic foamed plate processing equipment. The calender comprises a four-roller calendering unit, a laser ranging unit, a servo adjusting unit and a control system. The plate thickness is detected in a non-contact real-time mode by a laser triangulation ranging sensor, and the servo motor is linked with the PLC control system to realize closed-loop dynamic adjustment of the roll gap. The calendering method adopts multi-channel symmetrical calendering of the upper roller group and the lower roller group and gradient temperature control, so that the plate sequentially completes surface skinning, internal foaming and thickness setting calendering, and the thickness uniformity and the surface quality are effectively improved.

The application relates to the technical field of building construction formwork monitoring, in particular to a high-formwork displacement monitoring system and a monitoring method, which comprise a collection controller, a laser emitter and a CCD sensor connected with the collection controller; a positioning sheet is installed on a monitored point; the laser emitter emits laser to the monitored point; light is projected to the positioning sheet to form a light spot; the light spot is projected to the CCD sensor; the CCD sensor transmits a signal to the collection controller and then transmits the signal to a main server of a smart construction site management platform through a total station communication module; the main server is connected with a database, a display large screen and an alarm module; the application adopts a laser triangulation imaging principle, laser emitting angles can be adjusted by construction personnel, the problem that a CCD device is not convenient to install and debug due to a small light receiving area is solved, laser lines emitted by the laser emitter are formed into light spots through the positioning sheet, and then the light spots are projected to the CCD sensor, so that the service life of the CCD sensor is prolonged.

The embodiment of the invention provides a lens for improving laser triangulation ranging precision. In one specific implementation mode, the lens comprises a first negative meniscus lens with the concave surface side facing the object side, a second negative meniscus lens with the concave surface side facing the object side and a biconvex positive lens which are sequentially arranged from the object side to the image side along the optical axis of the lens; and the first negative meniscus lens and the second negative meniscus lens are glued and fixed. According to the embodiment, the first negative meniscus lens and the second negative meniscus lens are glued and fixed. The first negative meniscus lens and the second negative meniscus lens are glued and fixed to obtain the doublet lens, the lens adopts the design of the doublet lens and the single achromatic biconvex lens, the spherical aberration and the edge distortion of imaging can be effectively reduced, the mass center drift of light spots is inhibited, the size of the light spots is reduced, and the imaging definition is improved; therefore, the measurement precision of the laser triangulation ranging method is better improved.

The invention relates to a testing device for performing multi-dimensional comprehensive mechanical property evaluation on a bone and an external fixing device thereof. The device can simulate the stress state of a skeleton and a fixing device of the skeleton under the conditions of walking, supporting and loading in a multi-degree-of-freedom mechanical environment, and realizes high-precision loading of axial stretching, compression and bending loads and measurement of comprehensive mechanical characteristics such as static and dynamic properties. The device comprises a mechanical platform, a driving system, a force sensing system and a displacement measuring system. The mechanical platform adopts a structural form that double sliding tables are orthogonally arranged, and the two sliding tables can independently or cooperatively apply axial force and lateral force to realize multi-dimensional loading; the driving system comprises a servo motor and a ball screw transmission mechanism, rotation of the servo motor is converted into linear motion of the sliding table through a ball screw, and accurate control over loads is achieved. The force sensing system is composed of two sets of resistance strain type S-shaped force sensors, axial and bending loads are measured respectively, and force-displacement-time synchronous data collection is achieved. The displacement measurement system adopts a double-laser triangulation ranging structure, and high-precision measurement of angular displacement and linear displacement is realized by measuring micro displacement and dip angle change of a sample section through a reflection marking piece; data acquisition and control can be realized by controlling an acquisition system.

The invention relates to the technical field of metering and testing, in particular to a synchronism calibration device and method for a differential double-path laser triangular displacement sensor. The technical problem that an existing laser triangular length measurement system lacks a portable and easy-to-use synchronism calibration device is solved. The invention discloses a synchronization calibration device for a differential double-laser triangular displacement sensor. The synchronization calibration device comprises a supporting seat, a motor mounted on the supporting seat and a motor control box used for controlling the motor, a turntable is mounted on an output shaft of the motor; the turntable surface is uniformly and equally divided into m blades with the same thickness, and the m blades alternately fluctuate up and down relative to a reference surface by taking a radial surface passing through the center of the turntable as the reference surface, so that the turntable is in a square wave shape in the circumferential direction; and m is an even number for equally dividing the whole circle.

The invention discloses an iris lesion geometric parameter measurement system based on laser triangulation, and particularly relates to the technical field of optical measurement and geometric calculation, and the system comprises a laser projection module which is used for forming an iris surface topography modulation optical pattern; the image acquisition module is used for acquiring a modulated laser pattern on the surface of the iris; the data processing module is used for processing the laser pattern image and extracting sub-pixel coordinates of the lesion area of the laser pattern; the three-dimensional coordinate construction module is used for converting the sub-pixel coordinates of the lesion area into three-dimensional space coordinates; and the geometric parameter measuring module is used for calculating geometric parameters of the lesion area. The iris lesion geometric parameter measurement system based on laser triangulation is constructed through the laser projection module, the image acquisition module, the data processing module, the three-dimensional coordinate construction module and the geometric parameter measurement module, and the limitation of the prior art in the aspects of measurement dimension, precision, efficiency and automation is solved.

The invention belongs to the technical field of wind power equipment maintenance, and discloses a wind power precision bearing backlash measurement system, which comprises a laser ranging module, a data processing module and an auxiliary assembly, and is characterized in that the laser ranging module enhances a laser triangulation method. The method has the following advantages: the measurement efficiency is improved, the single measurement time is shortened from 2 hours to 3 minutes, the downtime is greatly reduced, and the usability of the fan is improved; the data reliability is improved, an encrypted report containing a timestamp, a GPS coordinate and an operator ID is automatically generated, human errors are reduced, and data traceability is ensured; the maintenance cost is reduced, the bearing replacement period is prolonged from 5 years to 7 years, the annual maintenance cost of a single fan is reduced by 120,000 yuan, and sudden faults are avoided through preventive maintenance; the safety performance is improved, the risk that workers enter the gearbox in a traditional method is avoided, the injury probability of operators is reduced, and the industrial safety standard is met.

The utility model relates to the related technical field of paint spraying robots, in particular to an efficient and accurate paint spraying robot which comprises a rack, a paint spraying mechanism is arranged at one end of the rack, in the using process, the module is externally connected with a control unit, and the control unit is connected with an external computer through a circuit. The module is mainly used for accurately identifying and positioning a workpiece before paint spraying, the surface of the workpiece can be rapidly scanned through the laser triangulation sensor, and shape, size and position information of the workpiece can be obtained. Then, an advanced image processing algorithm is used for analyzing and processing the information, the paint spraying area and path are accurately determined, the innovation point of the module is that a laser triangulation sensor and an accurate positioning algorithm of the module are used, paint spraying accuracy and efficiency can be greatly improved, paint spraying errors and waste are reduced, and in addition, the system is easy to popularize and use. The module can automatically adjust the paint spraying pressure according to different paint spraying requirements and workpiece surface characteristics, and the paint spraying quality is greatly improved.

The invention relates to a welding seam laser point cloud data acquisition and preprocessing technology, and provides a method for integrating multi-source data fusion and an AI enhancement algorithm, which comprises the following steps: S1, laser triangulation ranging: projecting a light band to the surface of a welding seam through a line laser, capturing a distortion image by combining a high-resolution camera, and carrying out laser triangulation; displacement sensor compensation and adaptive focal length adjustment functions are integrated, and the surface three-dimensional morphology is dynamically reconstructed; s2, multi-modal denoising: matching an optimal filtering strategy based on frequency domain analysis, fusing a deep learning model and improved bilateral filtering, and improving the image quality in a complex noise scene; s3, sub-pixel-level track extraction: a non-local mean-Hessian matrix fusion algorithm is combined with multi-scale eigenvalue decomposition to realize high-precision positioning of the center of the light band; s4, curvature-driven outliers are removed, specifically, a local curvature self-adaptive threshold value method is combined with a graph optimization algorithm, and weld edge feature points are prevented from being mistakenly deleted; and S5, layered resampling: preferentially reserving high-precision region points based on curvature importance, and optimizing space coverage uniformity. According to the method, the defects in the aspects of environmental adaptability, noise suppression and feature integrity in the prior art are overcome, high-speed, high-precision and high-robustness welding seam point cloud collection and preprocessing are achieved, and the method is suitable for special-shaped welding seam detection under complex working conditions.

The optical path symmetry type differential multi-reflection laser displacement sensor comprises a laser module, a mirror I, a lens group I, a lens group II, a mirror II, a lens group III, a mirror III, a diaphragm I, a diaphragm II and a photoelectric detector; the laser beam emitted by the laser module is shot to the mirror I, is collimated after being reflected through the lens group I, and is perpendicularly irradiated to a measured object; after being diffusely reflected by the measured object, one reflected laser beam is converged through the receiving lens group II to form reference light, and the other reflected laser beam is converged through the lens group III to form measurement light; the reference light is reflected by the mirror II, is focused and imaged on the photoelectric detector through the diaphragm I; the measurement light is reflected by the mirror III, is focused and imaged on the photoelectric detector through the diaphragm II; the mirror II and the mirror III are arranged in parallel and symmetry. The present application realizes high-precision displacement measurement by using the laser triangulation method, effectively suppresses the error caused by laser jitter, and improves the measurement stability.

The invention discloses a rolling automobile anti-collision beam corner processing automation device, and relates to the technical field of automobile anti-collision beam corner processing, the rolling automobile anti-collision beam corner processing automation device comprises a conveying line, the upper part of the conveying line is provided with a front end working condition pre-sensing module, the side surface of the conveying line is provided with a feeding mechanical arm, and the end part of the front end working condition pre-sensing module is provided with a three-axis adjusting mechanism. According to the automatic feeding device, automatic and accurate feeding of the anti-collision beams is achieved through the feeding mechanical arm, stable circulation of the conveying line is matched, and the manual intervention cost is greatly reduced; the front-end working condition pre-sensing module is used for accurately acquiring data of front-end rolling equipment and position information of a welding seam through a built-in industrial Ethernet communication module and a built-in laser positioning sensor, and is combined with three uniformly distributed laser triangular thickness measuring sensors and three ultrasonic hardness detectors to detect the thickness of burrs and the hardness of a base material in real time; and comprehensive data support is provided for the AI self-adaptive parameter regulation and control unit after preprocessing of the edge calculation module, and the problem of blind machining of traditional equipment is thoroughly solved.

This invention discloses a range-adjustable laser triangulation ranging sensor, comprising a transmitting unit and a receiving unit. The transmitting unit includes a first laser for emitting laser light and a first transmitting lens for focusing the emitted laser light. The first laser emits laser light, which is focused by the first transmitting lens and reaches the surface to be measured in a first range. After reflection, the laser light reaches the receiving unit. The transmitting unit also includes a second laser, a third laser, and an Nth laser, where N is greater than or equal to 2. The laser light emitted by the second, third, or Nth laser is reflected by a semi-transparent mirror to overlap with the laser emission path of the first laser, reaching the surface to be measured in the second, third, or Nth range, respectively. This ranging sensor, by incorporating multiple laser emitters within a single sensor, with each laser generator corresponding to a specific optical path and a specific range, enables range adjustment for a single sensor.

The application discloses a high-precision measurement device and a measurement method for five-degree-of-freedom error motion of a rotating shaft system, and comprises a rotating shaft system axial error motion measurement device, a rotating shaft system pitch error motion measurement device and a rotating shaft system radial error motion measurement device. The application realizes measurement of error motion of the rotating shaft system in the axial direction through an oblique laser triangulation method and a blade diffraction phenomenon, realizes measurement of error motion of the rotating shaft system in the pitch direction through a self-collimation principle and a small-hole diffraction phenomenon, realizes measurement of error motion of the rotating shaft system in the radial direction through a curved surface edge diffraction phenomenon, and finally realizes high-precision measurement of five-degree-of-freedom error motion of the rotating shaft system.

The invention discloses an integrated cleaning and detecting device and method for a marine steel wire rope. The integrated cleaning and detecting device mainly comprises a lower guide mechanism, a cleaning module, an upper guide mechanism, a drying module, a first middle guide mechanism, a three-dimensional magnetic flux vector sensing module, a second middle guide mechanism, a three-dimensional contour inspection module and an outer guide mechanism which are sequentially arranged. The cleaning module adopts a composite process of pre-coating a cleaning agent and combining high-pressure water jet with an annular rolling brush; the drying module rapidly dries the rope body through the alternate effect of hot air and cold air. The three-dimensional magnetic flux vector sensing module utilizes a permanent magnet to saturate and magnetize a steel wire rope, and adopts a three-axis TMR sensor array distributed in the circumferential direction to capture a three-dimensional vector magnetic field. The three-dimensional contour inspection module reconstructs a complete three-dimensional surface contour of the steel wire rope in real time based on a laser triangulation method. According to the integrated cleaning and detecting device and method for the marine steel wire rope, full automation of cleaning, drying and detecting is achieved, and the maintenance efficiency and reliability of the steel wire rope are remarkably improved.

The intelligent apple slice fine sorting and drying equipment comprises a conveying mechanism, a height limiting mechanism is mounted at the feeding end of the conveying mechanism, the conveying mechanism comprises a mounting side plate, and a conveying belt is mounted on the inner side of the mounting side plate; a plurality of protruding strip-shaped structures are arranged on the outer surface of the conveying belt at equal intervals. The intelligent selection area comprises a multispectral imaging mechanism, a color grading mechanism and a shape screening mechanism. Multi-spectral fusion detection: RGB + NIR + HDR imaging is combined to improve a YOLOv7 model, the defect detection rate is improved, the error rejection rate is reduced, dynamic color grading, HSV and spectral feature splicing + FCM clustering are adopted, incremental learning is supported, variety differences are adapted, the color sorting accuracy is improved, and the color sorting efficiency is improved. And meanwhile, 3D shape screening, laser triangulation and Poisson curved surface reconstruction are adopted, so that the warping degree detection precision is improved, and the rejection rate of deformed pieces is increased.

The present invention relates to a method for sorting out foreign bodies from a mass flow of sugar beets, the mass flow being transported through a detection unit by means of a conveying unit having at least one conveyor belt, a sorting unit being arranged downstream of the conveying unit, and the sorting unit being controlled by means of the detection unit and being configured such that foreign bodies are diverted out of an ordinary trajectory of the mass flow, wherein the detection unit carries out at least the following steps: I. in a first method step, objects in the mass flow arranged on the conveying unit are detected by means of laser distance measurement, in particular laser triangulation; II. in a second method step following the first method step, the detected objects are optically captured at least in part by means of hyperspectral or multispectral imaging, with first pixels being generated; III. in a third method step following the second method step, the generated first pixels are evaluated in order to detect foreign bodies in the mass flow; and in a fourth method step following the third method step, the sorting unit is controlled by the detection unit such that detected foreign bodies are diverted out of the ordinary trajectory of the mass flow.

Techniques for inspecting a substrate can include scanning the substrate across a plurality of scan positions using an interrogation beam having a width substantially equal to or greater than a diameter of at least one surface feature on a surface of the substrate to generate scanning results. A set of raw centroid calculation results can be generated based on the scanning results. An error property can be determined and removed from the raw centroid calculation results to generated refined centroid calculation results. A characteristic of the surface feature can be determined based on the centroid calculation results.

The invention provides a handheld laser welding gun with the focusing distance self-adaptive adjustment function. The handheld laser welding gun comprises an emitting part and an adjusting part. The emitting part comprises a first shell and a laser generator, an irradiation cavity is formed in the first shell, the laser generator is installed in an installation assembly in the cavity, and a laser beam of the laser generator penetrates out of the irradiation end of the first shell; the adjusting part comprises a second shell and a focus lens, the first shell is connected with the second shell, and the laser beam penetrates through the transmitting channel; the adjusting cavity is internally provided with an adjusting assembly for driving the focus lens to axially move along the transmitting channel so as to change the focusing distance, and the focus lens is arranged on the adjusting assembly; a laser triangulation distance measuring sensor, an infrared temperature sensor and a control part for receiving signals and controlling the adjusting assembly are arranged at the transmitting end of the second shell, a detachable grip is arranged at the bottom of the first shell, and a trigger part for controlling the laser generator to start and stop is arranged on the grip. According to the welding gun, self-adaptive adjustment of the focusing distance is achieved, and the problem that a traditional welding gun is poor in adaptability is solved.

The invention provides a speckle suppression method, and belongs to the technical field of measurement technologies and instruments. Obtaining a gray scale picture of the light spot through a laser triangular sensor; converting the grayscale picture into a numerical matrix; dividing a light spot area into a plurality of sub-areas with the same size based on the numerical matrix; calculating the center of mass of each sub-region in the direction; calculating an overall mean value and a standard deviation based on the centroid; sub-regions meeting conditions are screened out through a preset threshold value to form a set; and calculating the actual centroid of the light spot based on the sub-region set satisfying the condition. According to the method, a power weighting mechanism is introduced to enhance the contribution of high-intensity pixels to a centroid calculation result, and meanwhile, the interference of low-intensity noise pixels is inhibited, so that the calculated light spot centroid position is closer to a real center, and the measurement uncertainty caused by low-signal pixels is effectively reduced.

The application discloses an elevator traction sheave wear laser measurement method based on artificial intelligence, and realizes zero target, high speed and high-precision wheel groove wear detection through cooperation of a three-level neural network of a point cloud layer, a parameter layer and a diagnosis layer and laser triangulation, laser emits a radial fan-shaped laser surface, an encoder synchronously outputs angle information, and whole-circle pixel point clouds are collected; after coordinate conversion, the point clouds are solved by least square iteration to obtain a traction sheave pitch circle radius as a reference; the point cloud layer is denoised by a K-nearest graph + graph edge convolution network and outputs a pixel-level mask; the parameter layer predicts pulse intervals by two-layer LSTM, and the angle is corrected by Kalman filtering; the diagnosis layer shares features by five-layer convolution, three detection heads respectively output maximum wear depth and eccentric wear index, and the results of the first two heads are spliced with the pitch circle radius and elevator rated parameters, and the third detection head is used to regress the remaining life.

The application discloses a gravure roller processing defect detection method, and relates to the technical field of roller defect detection.The method comprises the following steps: step one, synchronously collecting multi-source data of the surface of the roller by a multi-modal sensor array and controlling the multi-source data, wherein the multi-modal sensor array comprises a linear array CCD, a laser triangulation sensor, an eddy current sensor and an infrared thermal imager.In the application, multi-modal sensor array is used to synchronously collect and fuse data, a three-dimensional integrated digital model containing texture, topography and physical characteristics is constructed, and a cascade deep learning network is combined for identification, so that closed-loop repair is finally realized.The application improves the comprehensiveness and accuracy of detection, can simultaneously find various macroscopic and microscopic defects on the surface and near the surface, greatly reduces the missed detection rate, enhances the automation and intelligent level, realizes full-process automation from detection, analysis to repair decision, and reduces the dependence on manpower.

The invention relates to the technical field of meat quality detection, in particular to a multimodal chicken breast lignification detection method based on a hyperspectrum and an image. The method comprises the following steps: shooting chicken breast by using an industrial camera to obtain an RGB image; the scanning hyperspectral imaging system scans chicken breast to obtain a hyperspectral image; scanning the chicken breast by using a laser triangulation system to obtain a three-dimensional laser image; carrying out correction and registration processing on the RGB image, the hyperspectral image and the three-dimensional laser image; extracting fusion features; and outputting a chicken breast lignification detection result through the deep learning model. According to the method, the RGB image, the hyperspectral image and the three-dimensional laser image of the chicken breast are acquired at the same time, the multi-modal deep learning model is constructed for joint discrimination, complementarity among different modal information is fully utilized, high-precision, objective and stable detection of the lignification state of the chicken breast is realized, and the detection accuracy is improved. The problem of insufficient information caused by a single detection means is avoided, and the risk of misjudgment and missed judgment is effectively reduced.