195results about How to "Realize high-precision detection" patented technology

The invention discloses a multi-modal data fusion-based high dam underwater defect measurement method, which comprises the following steps of: obtaining water flow velocity data, water pressure data and water quality turbidity data, performing preprocessing by adopting a self-adaptive median filtering algorithm, and constructing a hydrodynamic field feature model; then generating an initial trajectory based on the hydrodynamic field feature model and detection target area data, and performing trajectory optimization and stability evaluation in combination with dynamic constraint conditions; adjusting sonar parameters according to the turbidity influence and the trajectory parameters to perform data acquisition; carrying out space-time alignment through extended Kalman filtering, and carrying out feature fusion and optimization by adopting an attention mechanism; and finally, performing multi-scale consistency verification and accurate positioning mapping to obtain a final defect detection result. According to the invention, the problems of defect detection precision and reliability in a complex underwater environment are solved.

The invention discloses a casting product appearance defect detection method and system based on machine vision, and particularly relates to the technical field of casting product defect detection. According to the method, casting materials of a production line are divided into different categories, the casting materials with abnormal absorptivity are identified, terahertz wave scanning is carried out on the casting materials, scattering images are obtained, and the stability of scattering point density distribution is evaluated; scanning by using a terahertz time-domain spectrometer, analyzing an amplitude fluctuation state of a time-domain signal and evaluating a waveform distortion degree; after comprehensive analysis, the quality of images captured by the imaging system is evaluated, the images are divided into different grades, defect detection is directly carried out on high-quality images, the quality images are centered, the accuracy of the images is analyzed, and the transmitting power and the scanning speed of terahertz imaging equipment are dynamically adjusted, so that the imaging quality is improved, and the accuracy and the efficiency of detection are remarkably improved; and the quality and safety of casting products are ensured, and the product quality problem and potential equipment faults caused by untimely defect detection are effectively avoided.

The invention relates to an emotion recognition method based on a millimeter-wave radar and a robot. The method comprises the following steps: receiving a transmitting signal of the millimeter-wave radar and a corresponding echo signal, converting the transmitting signal and the corresponding echo signal into a point cloud format, and removing static reflection to obtain dynamic point cloud data and a timestamp of a surrounding environment; grouping the dynamic point cloud data into clusters; obtaining a centroid estimation value of the target, and tracking the centroid estimation value; sending the millimeter wave signal which horizontally moves at a constant speed to a target person, and receiving echo signals which are reflected by the target person and are uniformly spaced; point spread function optimization is carried out, and SAR image reconstruction is carried out by using a fast frequency domain near-field imaging method; calculating to obtain phase change of the human body signal, and separating to obtain respiration and heartbeat signals of the target person; and extracting emotion related features to obtain the current emotion of the target person. Compared with the prior art, the specific position of the human body can be accurately recognized, and non-contact privacy protection emotion recognition can be remotely achieved.

The invention provides a patient anesthesia state analysis method and system based on physical sign data, and relates to the technical field of anesthesia state monitoring and analysis. The system comprises an anesthesia state monitoring module, an electroencephalogram sign correlation analysis module, an anesthesia state local anomaly detection module and an anesthesia state diffusion detection module. According to the method, the correlation between the electroencephalogram state and the physical sign data is utilized, the anesthesia state of the patient is comprehensively analyzed by combining the electroencephalogram state and the physical sign parameters, static analysis of the local anesthesia stage and cross-stage dynamic diffusion detection are combined, a more refined and intelligent anesthesia state monitoring scheme is provided, and the monitoring accuracy of the anesthesia state is improved. The accuracy of anesthesia abnormity detection is improved, comprehensive monitoring and real-time early warning of the anesthesia state can be achieved, and the safety of a patient in the anesthesia process is ensured.

The invention belongs to the technical field of visual inspection, and particularly relates to a PCB defect detection method and device based on 2D and 3D image visual combination, and the method comprises the steps: collecting the 2D information and 3D information of a PCB, and carrying out the defect marking; the 2D information subjected to defect labeling serves as a training set to be input into a pre-constructed yolov10 target detection network, and a yolov10 model is obtained after training is conducted through a training strategy of double label distribution and consistent matching measurement; converting the 3D information subjected to defect marking into voxel features and coordinates by using a pre-constructed seond target detection network, performing feature extraction on the voxel features and coordinates by using a sparse convolutional neural network, and classifying object directions to obtain a trained seond model; and determining the defect type of the target PCB according to the yov10 model and the seond model. By adopting the deep learning algorithm and the 3D vision technology, the detection accuracy and efficiency can be greatly improved, the maintenance cost is reduced, and the detection robustness and stability are improved.

The invention discloses a spaceborne hyperspectral detection laser radar atmospheric particle inversion method. The method comprises the following steps of solving a laser ranging value and an altitude corresponding to a spaceborne hyperspectral resolution laser radar (HSRL) echo signal sampling point; reducing the noise of the echo signal by using a noise removal method; carrying out distance and system constant correction on the de-noised signal; a forecasting model is used for reanalyzing the data set, and S6, a molecular scattering signal transmittance lookup table is established through an atmosphere model and an iodine molecule hyperspectral filter absorption spectrum; calculating back scattering coefficients and scattering ratios of the cloud and the aerosol; performing hierarchical identification by using a scattering ratio and voltage signal-to-noise ratio threshold method; determining an effective inversion region through hierarchical region processing; calculating an initial extinction coefficient to a radar ratio; and performing Gaussian smoothing filtering on the radar ratio to obtain a final radar ratio, and calculating a final extinction coefficient profile. The method is characterized in that cloud and aerosol level identification is carried out on spaceborne HSRL echo signals, a molecular transmittance lookup table of a hyperspectral channel is established, and cloud and aerosol optical parameters are accurately and efficiently solved.

The invention relates to the technical field of bridge pile detection, and provides a bridge pile foundation defect detection system based on an ultrasonic technology. The control terminal is used for controlling and managing the whole operation process of the system and monitoring data in real time; the multi-channel high-frequency ultrasonic detection module is used for transmitting ultrasonic signals through a multi-channel high-frequency probe array and receiving reflection signals in the bridge pile foundation to form original detection data; and the parallel computing and 3D imaging module is used for carrying out parallel computing processing on the received ultrasonic signal data and generating a three-dimensional structure image in the bridge pile foundation. The multi-channel high-frequency ultrasonic detection module is combined with the phased array technology and the multi-reflection wave processing technology, and high-precision detection of the internal structure of the bridge pile foundation is achieved. The phased array technology can flexibly control the propagation direction and depth of ultrasonic waves, and the coverage of a detection range and the resolution of signals are remarkably improved, so that the system can detect hidden defects such as micro cracks and holes.

The invention relates to the technical field of stamping part surface detection, in particular to a stamping part surface detection method and a welding slag cleaning device.The detection method is based on three-dimensional point cloud data processing, is not affected by environment light changes, can more accurately separate a defect area from a background, and is high in detection accuracy. The welding slag can be effectively identified even at a complex geometrical shape or a deep concave position, comprehensive detection of a stamping part with a complex shape can be realized, and the accuracy and the efficiency of detection are remarkably improved; by presetting the basic threshold value and dynamically adjusting according to the reflection characteristics, it can be guaranteed that the stamping part can obtain stable detection and cleaning results under the detection conditions of different light environments.

The invention discloses a GIS internal humidity detection method and system based on a microsensor, and relates to GIS internal humidity detection.The GIS internal humidity detection method comprises the steps that the microsensor is adopted to record GIS internal environment data and initial calibration parameters, data preprocessing is conducted on the GIS internal environment data, and the GIS internal environment data comprises GIS real-time environment data and GIS historical environment data; constructing a dynamic environment compensation model according to the GIS internal environment data, and inputting the GIS real-time environment data into the dynamic environment compensation model to obtain dynamic environment compensation parameters; and training the initial calibration parameter to obtain an adjusted calibration parameter, feeding back the adjusted calibration parameter to the microsensor, updating the working state of the sensor, and performing detection again to obtain the internal humidity of the GIS. According to the invention, the accuracy and reliability of the humidity detection result are ensured, and the operation efficiency and safety of the GIS equipment are improved.

The invention discloses an all-optical ultrasonic imaging system based on phase modulation, and belongs to the technical field of imaging. The system comprises a laser assembly, an ultrasonic detector, an interference device and an imaging assembly, the laser assembly is used for providing continuous laser for the interference device; the interference device comprises a reference optical fiber arm, a sample optical fiber arm and an optical path adjusting device and is located between the ultrasonic detector and the imaging assembly. Wherein the interference device is used for dividing laser into reference light and detection light, and transmitting the detection light to the ultrasonic detector for phase modulation; the optical path adjusting device is used for adjusting the optical path difference of the two arms, forming interference light of the reference light and the phase-modulated detection light and inputting the interference light into the imaging assembly; the ultrasonic detector is used for generating ultrasonic waves, projecting the ultrasonic waves to a to-be-detected area, and performing phase modulation on the detection light in response to ultrasonic echoes of the to-be-detected area; and the imaging assembly is used for generating a corresponding image according to the interference light.

The invention discloses a remote sensing target detection method based on multi-scale feature collaboration and semantic complementation, and the method comprises the steps: S1, extracting the features of a given input remote sensing image through employing a backbone network, and generating a multi-level feature map after down-sampling; s2, collecting and mining global semantic information in the multi-level feature map; s3, combining the global multi-scale features and the local hierarchical features, and fusing semantic and spatial context information; s4, fusing significant information of adjacent layers in the input feature map to enhance semantic representation of the output feature map; s5, the classification and positioning decoupling detection head predicts the category and the detection frame of the target from the enhanced feature map; and S6, selecting Distribrute Focus Loss and Complete IoU Loss as regression loss to be used for bounding box regression of the target, wherein the Distribrute Focus Loss and the Complete IoU Loss are used as the regression loss. The method provided by the invention is beneficial to modeling the relationship between the object and the background in the remote sensing image, reduces the influence of the complex background on the accurate target characteristics, adapts to the large-scale change of different remote sensing targets, and improves the detection accuracy.

The invention belongs to the technical field of crimping tube X-ray detection. The invention provides a power transmission line crimping pipe X-ray detection system and an autonomous detection method. A laser radar and a binocular vision sensor are both in communication connection with a control terminal; the robot body comprises a walking rack, an X-ray detection mechanism fixed on the walking rack and a driving wheel located at the bottom of the walking rack, the walking rack, the X-ray detection mechanism and the driving wheel are both of a wave-shaped structure, and the wave-shaped structure at least comprises two concave parts; the driving wheels are driven by a driving motor, the control terminal is in communication connection with the driving motor to control the robot body to move forwards or backwards, and the control terminal is used for conducting navigation control over the X-ray robot according to data of the laser radar and the binocular vision sensor; the detection precision and comprehensiveness of the crimping pipe with the reinforcing wire are improved, and the walking stability of the robot on the power transmission line is improved.

The invention relates to the technical field of parking lot management, in particular to an intelligent parking management system and method based on the Internet of Things, and provides the following scheme that a parking space image is obtained through an image sensor, the parking space occupancy state is judged through the neural network technology, and the optimal parking space is recommended according to the parking preference of a vehicle owner and the real-time parking space state. Meanwhile, the system adjusts the parking fee in real time according to the parking space occupancy rate of the parking lot, the traffic flow and the parking duration of the vehicle owner through a dynamic pricing strategy. The monitoring module can remotely monitor the parking lot, discover and process abnormal conditions in time, and improve the efficiency and safety of parking management.

The invention discloses a method for detecting track bolt defects based on YOLOv8, and relates to the technical field of bolt defect detection, and the method comprises the following steps: S1, collecting a track bolt sample image, and preprocessing the image; step S2, inputting the image after expansion processing into an improved YOLOv8 detection model; firstly, multi-size feature fusion is carried out on an input image, and a multi-scale output track bolt local feature map is obtained after fusion; clustering the object sizes in the track bolt data set by adopting an improved K-Means + + clustering mode to obtain the optimal priori frame size of the track bolt data set; then utilizing information of a prior frame, mapping tx and ty between 0 and 1 by adopting a sigmoid function based on an improved YOLOv8 detection model in a central coordinate conversion process of a prediction frame, and adding grid unit coordinate information to obtain final prediction frame information; and finally calculating the difference between the prediction frame and the real frame by using a loss function.

The invention discloses a battery performance detection system based on big data analysis, which relates to the technical field of battery performance detection and comprises a data acquisition module, a data transmission module, a battery analysis module and a user interaction module. The method can plan the replacement or maintenance plan of the battery in advance, optimizes the operation cost and reliability of energy storage, can track the degradation process of the battery performance along with time and use conditions, and helps an enterprise to know the performance of a battery product in actual use in advance. Meanwhile, the product design and the production process can be timely adjusted for enterprises, and the product quality is improved; for a user, the use and maintenance of the battery can be better arranged, and the effective service cycle of the battery is prolonged.

The invention discloses a gearbox component surface defect detection method and system, and the method comprises the steps: obtaining a gearbox component surface image through a camera device, and carrying out the graying processing; performing segmentation of a gearbox component entity image on the grayscale image through a first U-net model to generate a binary image identifying a gearbox component entity, and performing Hadamard product operation on the binary image and the grayscale image to generate a construction image; performing defect area image segmentation on the constructed image through a second U-net model to generate a binary image for identifying a surface defect area of the gearbox component; calculating the number of non-zero pixels in the binary image for identifying the surface defect area of the gearbox component and the number of non-zero pixels in the construction image, and calculating the qualified confidence coefficient of the gearbox component according to the number of non-zero pixels in the binary image and the number of non-zero pixels in the construction image; and judging whether the gearbox component is qualified or not according to the qualified confidence coefficient of the gearbox component. The device is simple in structure, does not consume too much computing resources, and can achieve high detection precision without a large number of training samples.

The invention discloses a coal mine shaft crack identification method in a non-illumination environment based on an optical fiber sensing grating, and belongs to the field of coal mine shaft disease detection. According to the method, distributed optical fibers form an optical fiber sensing grating, the optical fiber sensing grating is fixed on the surface of a well wall, an optical fiber demodulator is used for collecting strain along the optical fibers, a blank blind area of the optical fiber sensing grating is filled through an interpolation algorithm, and a strain field image data set of the well wall is obtained; and a CCD industrial camera is matched to shoot a well wall crack image sample at the optical fiber sensing grating, binarization processing is carried out, and a well wall crack image data set is established. And establishing a well wall crack identification model based on deep learning, introducing two parallel feature extraction branches, respectively receiving an optical fiber strain field image data set and a well wall crack image data set, and performing feature extraction and fusion of the two different types of data sets through training. Because the strain data is sensed through the optical fiber, the illumination environment is not needed, and after the model training is completed, the identification and early warning of the well wall crack in the illumination-free environment can be realized only by inputting the optical fiber strain data acquired in real time into the trained model.

The invention provides a photovoltaic module detection device and detection method. Relates to the technical field of photovoltaic structures. The photovoltaic module detection device comprises a first transmission assembly and a second transmission assembly which are arranged continuously. The first transmission assembly is provided with a movable first image acquisition unit, and the first image acquisition unit can move along a first direction and / or a second direction relative to the first transmission assembly so as to acquire a first image of a specified area on the photovoltaic assembly and identify and judge defect parameters on the area according to the first image; a movable second image acquisition unit is arranged on the second transmission assembly; and the second image acquisition unit can move along the first direction and / or the second direction relative to the second transmission assembly so as to acquire a second image of other areas except the specified area on the photovoltaic assembly and identify and judge defect parameters on the area according to the second image.

The invention belongs to the technical field of space environment detection, and relates to an ASIC (application specific integrated circuit)-based multichannel parallel charge detection device for space charged particle detection, which comprises an anode plate, a front placing plate and a digital acquisition and processing plate, and signal transmission is performed between the anode plate and the front placing plate as well as between the front placing plate and the digital acquisition and processing plate through inter-plate connectors; the anode plate is an annular anode and is divided into a plurality of anodes, the output of each anode is connected with a current-limiting resistor in series, and a charge signal output by each anode is transmitted to the front amplifier plate through the inter-plate connector; the pre-amplification board adopts an integrated charge sensitive amplifier ASIC to amplify multiple paths of charge signals output by the anode plate, and outputs multiple paths of digital pulses to be input into the digital acquisition processing board through an inter-board connector; and the digital acquisition and processing board is used for detecting and counting multiple paths of digital pulses output by the pre-amplification board in parallel, and periodically packaging and outputting the digital pulses to a PC (Personal Computer) end.

The invention relates to a precision detection device for gear processing, which is suitable for detecting the dynamic performance and single tooth precision of a gear, and comprises a base, a rotating bracket, a fixed block, a motor, a reduction gearbox, a test gear, a to-be-tested gear, a main measuring rod, a plurality of measuring rods and the like. High-precision detection of gear dynamic balance, vibration parameters and the tooth outer end and the contour line of a single gear is achieved, the reduction gearbox reduces the rotating speed of the gear and strengthens torque control, and the gear thickness difference design is matched, so that the multiple measuring rods can slide along the tooth outer end of the gear, tooth surface deformation and the contact state are accurately detected, and the device is made of high-carbon low-alloy tool steel. The gear detection device has high strength and wear resistance, adapts to various gear specifications, can output detection data and analysis results in real time, is simple to operate, high in detection precision and high in applicability, and can be widely applied to the fields of gear manufacturing, quality detection and maintenance.

The invention provides a fastener concave-convex amount measuring method based on monocular surface structured light. The fastener concave-convex amount measuring method comprises the steps that internal parameters of a monocular camera and a projector are determined; calculating the relative position of the monocular camera and the projector according to the internal parameters of the monocular camera and the projector; in the relative position, the projector projects a grating pattern to the fastener surface, while the monocular camera captures a second image of the grating pattern; detecting a fastener area from the second image; calculating an absolute phase value of the third image of the fastener region; according to the internal reference of the monocular camera and the projector, the relative position of the monocular camera and the projector, and the absolute phase value of a third image of the fastener area, converting the absolute phase value of each pixel in the third image into coordinates of the object surface in a three-dimensional space by using a phase-height mapping relation, and obtaining a three-dimensional point cloud; and calculating the concave-convex amount of the fastener based on the three-dimensional point cloud.

The invention discloses a pellet positioning circle measurement defect detection method and device based on an SOD-YOLOv5 model, and the method comprises the following steps: placing a plurality of pellets in grid holes of a pellet box, precisely positioning the center of each pellet through a microscope lens, detecting a pellet image through the SOD-YOLOv5 model, calculating the center offset, and determining the defect of the pellet positioning circle measurement defect. Fine positioning of the pellet is realized; then, the maximum inscribed circle of the pellet is extracted, edge detection is carried out along the circumference, and a circular contour is fitted through a least square method so as to calculate geometrical characteristics such as roundness and diameter; identifying and positioning the surface defects of the pellet by using an SOD-YOLOv5 model and a difference method; a microscope lens automatically moves to the position of each hole, and multi-pellet detection is carried out in sequence; the device can adapt to various colors and illumination conditions, high-precision, automatic and batch detection is realized, and the reliability of a detection result is ensured.

The invention discloses an additive manufacturing quality detection system and method based on thermal imaging. The invention relates to the technical field of image processing, and solves the problems of low detection accuracy and low real-time performance in the existing additive manufacturing quality detection field. The additive manufacturing quality detection system based on thermal imaging comprises a model printing module, a data acquisition module, a data processing module and a data comparison module, the model printing module comprises a powder spreading rod, a powder feeding bin, a forming bin, a platform and a scanning galvanometer; the quality of each printing layer can be comprehensively evaluated in real time, the detection quality and the printing efficiency are improved, and real-time detection of the printing quality in the additive manufacturing process is achieved.

A landfill and leachate distribution range detection method based on multiple physical property parameters relates to the field of environmental protection, and comprises the following steps: acquiring a target area selected by a user, and detecting the target area by adopting a multi-frequency electromagnetic method to obtain surface area detection data; determining a preliminary distribution range of landfill and leachate according to the surface area detection data; according to the initial distribution range and the surface area detection data, determining a wiring area of refined detection; detecting electrical structure parameters of different depths in the wiring area by adopting an electrical sounding method, and generating line area detection data; determining a plurality of detection point locations according to the wiring area and the line area detection data; detecting speed construction parameters of different depths in the detection point by adopting a micro-motion method, and generating point area detection data; and inputting the surface area detection data, the line area detection data and the point area detection data into a joint inversion formula to obtain a distribution range result. By implementing the application, the landfill and the percolate in the designated area can be accurately detected.

The invention discloses a terahertz transmission-based plastic pipe wall thickness testing method and device, and relates to the technical field of terahertz transmission, and the method comprises the following steps: constructing a standard model; setting the frequency of standard terahertz waves and the width of wave beams, and constructing a standard setting array to obtain a plurality of to-be-screened setting arrays; the terahertz emission source emits a plurality of terahertz waves under a set array to be screened to obtain a plurality of thickness distribution diagrams with different fineness degrees; acquiring signal-to-noise ratio influence data and measurement precision data when the wall thickness of the plastic pipe is tested; constructing a thickness distribution diagram resolution evaluation model based on a long-short-term neural network; carrying out data analysis to obtain a screened set array, and applying the screened set array to the wall thickness test of the plastic pipe; according to the invention, the problem that the final test effect is affected because the fine thickness change cannot be accurately captured while the signal intensity is ensured because the resolution setting of the thickness distribution diagram is not standard is solved.

The invention relates to a photo-thermal microscopic imaging system and method based on a cyclic generative adversarial neural network, and belongs to the technical field of photo-thermal science and detection and signal processing. Comprising a reflector, a sleeve lens, a first slide, a scanning lens, a platform, a galvanometer controller, a dichroscope, a polarization cube, a reflector group, a second slide, a third slide, a spatial light filter, an acousto-optic modulator, a first optical filter, a polarization cube, a fourth slide, an exciting light source, a polarization cube, a spatial light filter and a second optical filter. A probe light source; and a photo-thermal microscopic imaging system. According to the method, the three-dimensional diffusion of the heat flow is fully utilized, and the defect is accurately reconstructed from a two-dimensional angle by changing different spatial fluctuation states of the heat flow, so that the detection depth and the detection efficiency are improved, and meanwhile, the quantitative characterization of the material and the three-dimensional chromatography reconstruction of the geometric structure can be realized; according to the method, efficient and accurate detection and three-dimensional feature reconstruction of superficial layer defects of composite materials, metal materials and high-molecular polymers are achieved.

The invention relates to the technical field of display detection, and more particularly discloses a display detection device and method based on visual inspection, and the device comprises a base assembly, the base assembly comprises a first base, the top of the first base is fixedly connected with a top plate, and the top of the first base is provided with a first middle groove and a second side groove; a transmission assembly is mounted in the first middle groove, the transmission assembly comprises a first motor and a first screw rod, a translation assembly is mounted on the outer side of the first screw rod, and a light source assembly and a detection assembly are mounted at the bottom of the top plate. Through the arrangement of the light source assembly and the translation assembly, the bottom of the light-emitting plate can emit parallel light beams to irradiate the display screen, the light is reflected by the surface of the display screen and absorbed by the detection assembly, the position of a drop point is calculated, and the thickness change value of the display screen is calculated to be converted and expanded into a distance difference; and high-precision detection of the thickness of the display screen is realized.

The invention discloses a high-precision self-adaptive frequency hopping signal detection method and device and computer equipment, and relates to the technical field of radio monitoring, and the method employs a mode of detecting pulses one by one through spectral lines, and can achieve the detection of frequency hopping signals of any frequency band and any rate, including a plurality of frequency hopping signals of different rates, which are aliasing together. In addition, related parameters of the frequency hopping signal can be accurately given, including signal strength of each hopping point, a frequency set of the frequency hopping signal, a frequency hopping rate, a duty ratio of the frequency hopping signal and the like. The method is basically not influenced by various interferences of electromagnetic environments, and high-precision detection under various electromagnetic environments is realized.

The invention discloses an AI-based PCBA circuit board defect detection system. The AI-based PCBA circuit board defect detection system comprises the following steps: S1, an image acquisition module, wherein the image acquisition module comprises a camera shooting assembly and a light source assembly; wherein the camera shooting assembly comprises a high-resolution camera, and the high-resolution camera is used for capturing an image of a PCBA circuit board to be detected; s2, an image processing module, wherein the image processing module comprises image preprocessing, image segmentation and image recognition; according to the invention, automatic defect detection is carried out on the PCBA circuit board by using an artificial intelligence technology, the system combines image processing and deep learning technologies to realize rapid and accurate detection of surface defects of the circuit board, and meanwhile, through an automatic and intelligent detection means, the detection precision, efficiency and consistency are greatly improved; and the system not only can meet the current production requirements, but also can be continuously optimized and expanded through data driving, and has remarkable competitive advantages.

The invention discloses a decoration firing and gilding detection method and device in the technical field of non-contact optical detection. The method comprises building a detection system; the method comprises the following steps: placing N glass bottle samples on a workpiece table, collecting multi-view images, performing morphological processing to obtain a sample set, and manually marking the images in the sample set; building a deep learning network, and inputting a sample set for training; performing multi-view imaging on the to-be-detected glass bottle, performing morphological processing on an imaging result, and inputting the imaging result into the deep learning network for initial detection; the similarity of decoration firing and gilding marking results is used as feedback input, the upper computer controls the workbench to adjust the glass bottle, and secondary imaging is carried out on the marking position; and performing fine detection on the secondarily imaged image after morphological processing, and outputting a detection result. In the application, the method has good automation capability, high-precision detection of decoration firing and gilding can be realized through secondary imaging, adverse effects of light intensity and spectrum can also be inhibited through morphological processing, and the detection precision is improved.