42results about How to "Solve missing detection" patented technology

The invention discloses a diseased fish detection method and device, and belongs to the technical field of computer vision detection. In order to solve the problems that in the prior art, diseased fish disease detection efficiency is low, accuracy is not high, and real-time detection cannot be achieved, a diseased fish detection model is constructed, and the diseased fish detection model comprises a backbone network, a neck network and a detection head network; the method comprises the following steps: inputting a diseased fish image into a backbone network based on a spatial deep convolutional neural network for feature extraction to obtain a visual feature map, and then inputting the visual feature map into a contrast-enhanced attention transfer structure of a neck network for feature fusion processing to obtain a fused feature map; and the fusion feature map is input into a detection head network, and a sick fish target detection result is obtained through detection of a multi-head prediction small target detection algorithm, so that fish school disease detection is performed in real time, the detection efficiency and detection precision of fish diseases are effectively improved, and the method has relatively high practicability and application universality.

The invention provides a track fastener state recognition method based on a three-section type parallel network architecture. The method comprises the steps that a track inspection image is acquired; constructing a convolutional neural network (CNN) backbone network, and performing multi-scale feature extraction on the image of the track fastener; performing convolution processing and up-sampling processing on the fastener features by using the local feature branches to obtain local fastener features; performing global average pooling and convolution processing on the fastener features by using the global feature branches; the method comprises the following steps of: processing the characteristics of the fastener by using a mask generation branch of a framework based on Vision Transform, and generating a prototype mask of the track fastener; local, global and mask fastener features are fused; performing non-maximum suppression and mask generation on the fastener feature map; and analyzing the health state of the track fastener. According to the method, the CNN backbone network and the Vision Transform are combined, a new three-section parallel network architecture is provided, accurate identification and state analysis of the track fastener are realized through instance segmentation, and the identification accuracy in a complex environment is improved.

The invention discloses a lithium mineral component detection method based on improved YOLOv8, and belongs to the technical field of computer vision and mineral detection.The method comprises the steps that dry powder samples of various lithium mineral components are collected, the collected dry powder samples are shot under an optical microscope, and then the shot images are segmented and marked to obtain lithium mineral components; constructing a sample data set covering various lithium mineral components; the method comprises the following steps of: improving a YOLOv8 network to obtain an improved YOLOv8 network; training the improved YOLOv8 network on the basis of the sample data set; and the trained and improved YOLOv8 network is utilized to realize lithium mineral component detection. By adopting the scheme, the lithium mineral component detection effect and efficiency can be improved, and technical support is provided for intelligent detection of lithium ores.

The invention discloses a vehicle-mounted camera visibility inversion method and device, a medium and electronic equipment, and the method comprises the steps: obtaining an image frame data set shot by a vehicle-mounted camera, carrying out the preprocessing, and constructing a visibility inversion data set; a visibility inversion model is constructed, and the visibility inversion model takes the visibility true value as output and is obtained by training and optimizing network weight; and inputting a picture shot by a vehicle-mounted camera in real time into the optimized visibility inversion model to obtain a current visibility true value. According to the method, the corresponding visibility can be accurately estimated, and high-precision visibility real-time inversion is realized.

The invention relates to the technical field of coal mill powder pipes, in particular to a powder leakage defect automatic identification method based on digital twinning and AI video analysis, and the method comprises the steps: obtaining operation parameters and environmental parameters of a coal mill powder pipe, calculating a powder leakage risk characterization coefficient, and judging whether there is a potential risk of powder leakage; image data of the surface and the surrounding area of the powder pipe in the initial starting period are collected through an AI video analysis module, powder leakage characteristic parameters are extracted, a characteristic change curve segment is drawn, similarity comparison is conducted on the characteristic change curve segment and the reference characteristic curve segment, and whether a suspected powder leakage area exists or not is judged; actual operation parameters and model prediction parameters of the powder pipe in a stable operation period are obtained based on a digital twin model, the fluctuation degree of parameter deviation is calculated, and whether the powder leakage defect exists or not is comprehensively judged by combining a suspected powder leakage area judgment result; the early-stage accurate identification of the powder leakage defect is realized, the detection efficiency and precision are improved, and the operation and maintenance cost and the safety risk are reduced.

The invention discloses a textile defect detection method which comprises the following steps: step A, image acquisition: performing real-time and / or timed acquisition on a textile part of a textile needle by virtue of a visual sensing system to acquire images; step B, constructing a feature extraction network: based on a YOLOv5 backbone network, introducing a multi-branch discrete cosine attention mechanism module (MDCA) to form a new backbone network SCNet, and drawing an output feature map of the new backbone network SCNet according to the acquired image; and step C, adopting an MUPool pooling method: a new backbone network SCNet outputs a feature map, important node features are acquired and screened through multi-view map convolution, a cascade structure and a rear-end fusion module, and finally a multi-scale feature map is generated, so that the problems of false detection and missing detection caused by oil stains, black oil needles, bright oil needles and dirty yarns can be solved while relatively high detection precision is ensured.

The invention discloses a rape pod detection and counting method and system based on YOLOv8, and the method comprises the steps: constructing a rape inflorescence pod image data set, and carrying out the training and testing of the YOLOv8, and obtaining a detection model with stable and standard detection precision; inputting the detected rape inflorescence silique image into the detection model, and outputting a bounding box and confidence coefficient for calibrating the rape silique by the detection model; and sorting the bounding boxes, calculating an overlapping degree between two adjacent bounding boxes, screening the bounding boxes according to the overlapping degree, and counting the number of the bounding boxes reserved by screening to obtain the total number of rape inflorescence siliques. The detection model obtained by the method meets the detection requirements of small-target rape pods, the problems of missing detection and repeated counting are solved through bounding box screening, and the counting accuracy is high. According to the system, full-process automation of input-detection-counting-output is achieved, the operation threshold is lowered, and the actual use requirement of a user can be better met.

The present invention relates to a microbial immunochromogenic reagent for fluorescent staining of vaginal secretion wet mounts, in particular to a microbial immunochromogenic reagent and a preparation method and application thereof. The microbial immunochromogenic reagent of the present invention comprises a fluorescent stain, a buffer solution, a permeabilizing agent and an antibacterial agent. The microbial immunochromogenic reagent and the microbial sample are mixed, and wet mount detection can be performed under a fluorescence microscope. This operation is simple, fast and convenient, greatly reduces the workload of operators, and can achieve rapid detection of a large number of samples. It also has high sensitivity and accuracy, and a wide range of applications.

The present invention belongs to the technical field of target recognition, and provides a target detection method and system based on a multi-scale mixed dimensional network. The method comprises: inputting a target image into a trained multi-scale mixed dimensional network, removing redundant frames from the obtained image by an NMS method, and obtaining an image marked with a target position and category; the multi-scale mixed dimensional network comprises a pyramid network, and a mixed dimensional module is introduced between every two convolutional layers from top to bottom of the pyramid network.

The invention relates to a detection device and a detection method, and belongs to the technical field of circuit board quality detection.The detection device comprises a bottom plate, a bearing plate and a light transmission detector, and the bottom plate is provided with a weighing device and an electrostatic eliminator; the bearing plate is parallel to the bottom plate, and the bearing plate is connected with the bottom plate through the weighing device; the static electricity eliminator is annularly arranged on the side face of the bearing plate, and the ion output end of the static electricity eliminator is located on the datum plane of the bearing plate; and the light transmittance detector is positioned on the reference surface of the bearing plate, is positioned on one side of the bearing plate, and is used for detecting the light transmittance of the sheet. According to the scheme provided by the invention, the first adhesion entrainment judgment is carried out on the sucked sheet through the weighing sensor, and the second adhesion entrainment judgment is carried out on the sucked sheet through the light transmission detector, so that the adhesion entrainment detection accuracy is improved, and the problems of wrong detection and missing detection are solved.

The invention discloses an educational academic literature tracing method and system based on retrieval enhancement generation. The method comprises the following steps: performing fine-grained analysis on the multi-source literature, constructing a document object containing fields such as a title, an abstract, a methodology, an experimental result and a conclusion, and establishing a field-level index and a citation network feature score; identifying an academic intention queried by a user by utilizing a large model, dynamically loading a weight mapping table according to the academic intention, and performing weighted rearrangement on multiple paths of retrieval results to obtain candidate core literatures; generating a traceability text with an explicit label based on the candidate literature; and executing reference consistency verification by using the natural language inference model, and executing illusion correction according to a verification result. According to the method, through structured field analysis and dynamic rearrangement of intention driving, the problems of semantic fragmentation and reference fictition when a general RAG is used for processing complex academic literatures are solved, and the academic traceability preciseness and accuracy are remarkably improved.

The invention discloses a traffic sign detection method under a complex background based on multi-scale feature fusion. The method comprises the following steps: A1, inputting an image; a2, backbone network feature extraction: the image is processed by a DS-C3 module, and multi-scale feature information is extracted; a3, feature fusion: carrying out fusion processing on the extracted multi-scale features through a newly added 160 * 160 small target feature map, generating a feature layer containing high-precision information, and enriching the scale of feature extraction; a4, detecting and classifying; and A5, outputting a result. By introducing the DS-C3 module and using the depth separable convolution, the parameter quantity and the calculation quantity of the model are greatly reduced, the calculation efficiency is remarkably improved, and meanwhile, relatively high detection precision is kept; aiming at the problem of small target traffic sign detection, the detection performance of a small target is specially optimized. According to the design, the precision of the model in detecting the small traffic sign is effectively improved, and the common problems of missing detection and false detection in the prior art are solved.

The invention relates to the technical field of artificial intelligence image recognition and data mining, and discloses an image data mining processing system and method combining a visual large model and a visual small model. The method aims at solving the problems that in the prior art, the image recognition recall rate and accuracy are difficult to consider, the calculation cost and the processing speed are unbalanced, the data management efficiency is low, and the model generalization ability is insufficient. The system comprises a data input module, a frame extraction deduplication module, a picture library construction module, a visual large model judgment module, a visual small model judgment module, a label preprocessing module, a manual inspection module, a training data generation module, a model training module and a data security module. According to the method, the generalization ability is guaranteed through preliminary screening of an LPM (Deic) large model, missing detection is complemented through secondary verification in cooperation with a YOLO small model, false alarms are eliminated in combination with manual verification, meanwhile, double models are optimized through targeted sample training, the pain point of missing detection or misjudgment of a single model is effectively solved, and accurate and comprehensive consideration of target recognition is achieved.

The present application relates to a kind of VCSEL chip lens defect intelligent identification and diagnosis method and system, it is related to defect identification technical field, including the following steps, by polarized light source irradiation is carried out optical imaging to the VCSEL chip lens, obtains polarization reflection image;Based on the polarization reflection image carries out gray threshold segmentation, obtains independent lens area;To the independent lens area carries out multi-scale pixel matrix convolution, obtains local edge profile, and based on the local edge profile determines whether the VCSEL chip lens exists defect geometric feature;If there is, then based on the defect geometric feature carries out wafer physical coordinate mapping, obtains defect distribution atlas;Based on the defect distribution atlas is carried out defect type diagnosis to the VCSEL chip lens, generates defect type report, solves the technical problem that significant missed detection and misjudgment exist in micron level defect identification in traditional technology.

The invention discloses a traffic image target detection method based on YOLO-Light, belongs to the technical field of image recognition, and solves the problem that detection precision and target tracking stability are affected by missing detection and false detection which are easy to occur for a target which is small in size, complex in scene and frequent in shielding under the view angle of an unmanned aerial vehicle in the prior art. At the backbone of YOLO-Light, after feature extraction is carried out on the traffic image, feature extraction is carried out again through an improved fast space pooling layer, and a traffic feature map is output; the number of the last three layers of channels of the backbone of the YOLO-Light is adjusted; at the neck of the YOLO-Light, the traffic feature map is subjected to multi-cascade up-sampling, layer connection and feature decomposition, then multi-scale feature fusion is carried out, and a multi-scale traffic feature map is output; and at the head of the YOLO-Light, sequentially inputting a space attention mechanism and a detection head into the multi-scale traffic feature map to detect a traffic target.

The application discloses a kind of middle voltage distribution equipment grounding fault precision detection method, it is related to distribution network fault detection and protection technical field.The method includes: in each feeder head synchronous acquisition zero sequence current and transient traveling wave signal;Zero sequence current and transient traveling wave are normalized and fused to construct joint transient signal, and net energy ratio is calculated by wavelet packet decomposition;Equivalent dynamic grounding impedance and impedance homology change rate are calculated;Intermittent grounding coupling index is constructed by jointly identifying arc reignition;Multi-feeder correlation discriminant matrix is constructed, and fault feeder is positioned according to fault correlation discriminant value sorting.The application realizes the collaborative use of zero sequence quantity and high-frequency transient information, improves the detection accuracy and positioning reliability of complex grounding fault.

The invention relates to the technical field of panel detection, in particular to a hole edge microcrack defect detection method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining an original bright field image and an original dark field image of a to-be-detected liquid crystal screen hole edge under telecentric imaging; inputting the original bright field image and the original dark field image into an improved microcrack defect detection model to obtain a detection result of the hole edge of the liquid crystal screen to be detected; the method is combined with an improved microcrack defect detection model, efficient and accurate detection of the liquid crystal screen hole edge microcracks is achieved, the detection speed and accuracy are improved, and the false detection rate and the omission ratio are reduced.

The invention provides a coal logistics park traffic flow statistical system and method based on YOLOv8 and OpenCV, and the method comprises the following steps: S101, deploying high-definition monitoring equipment at a key intersection in a coal logistics park, and configuring equipment parameters; s102, variables are initialized, a model is loaded, and a data loader is configured; s103, acquiring camera video data in real time through the rtsp video stream; s104, traversing the data frame, and carrying out object detection and tracking; s105, processing a detection result to realize vehicle counting; and S106, deploying the model to a management end to display a result. According to the method, targeted training is carried out on the YOLOv8 model for the special environment of the coal logistics park, the OpenCV and the vehicle tracking algorithm are combined, efficient and accurate recognition, tracking and statistics of the vehicles are achieved, the detection problem caused by shielding, light change and other factors is solved, and a basis is provided for park traffic management. In addition, the system is flexible and expandable, a user can configure the system according to needs, the system can be conveniently expanded to other logistics parks or traffic scenes, and large-range vehicle monitoring is achieved.

The invention discloses a machine vision-based PVC plastic pipe fitting defect detection method and device, and relates to the technical field of PVC plastic pipes, and the method comprises the following steps: obtaining an image of a PVC plastic pipe fitting, setting an initial difference value area range based on the geometrical characteristics of a defect contour, and obtaining a defect contour of the PVC plastic pipe fitting; carrying out expansion operation and corrosion operation on the initial difference value area range for a plurality of times to obtain a plurality of adjustable difference value area ranges; performing defect detection; respectively acquiring defect positioning data and detection robustness data; constructing a difference value area range evaluation model; constructing a difference value area range-detection effect relation curve, performing data analysis to obtain an adjusted difference value area range, and applying the adjusted difference value area range to defect detection of the PvC plastic pipe fitting; the problems that in the prior art, difference value area range selection is not standard, a dynamic adjustment mechanism suitable for different defect types lacks, information missing or noise interference is caused, and the accuracy and applicability of defect detection are seriously restricted are solved.

The invention provides a fully mechanized coal mining face sensor fault diagnosis method and device and electronic equipment, and the method comprises the steps: obtaining the current working stage of a target hydraulic support, and obtaining the time period sensing data of a to-be-diagnosed sensor on the target hydraulic support at the current working stage; according to the current working stage, a target diagnosis path is determined, and the target diagnosis path is formed by combining one or more diagnosis algorithms; and obtaining a fault diagnosis result of the to-be-diagnosed sensor according to a target diagnosis path and the time period sensing data. According to the method, the adaptive target diagnosis path is dynamically selected according to different working stages of the target hydraulic support, and analysis is performed in combination with the corresponding time period sensing data, so that the problem of false alarm or missing detection caused by mismatching of working conditions in a traditional static diagnosis method is effectively solved, the accuracy and robustness of sensor fault diagnosis are remarkably improved, and the fault diagnosis efficiency is improved. The low false alarm rate, the low omission rate and the high operation and maintenance efficiency are achieved.

The application is suitable for the technical field of eye movement data processing, and provides a pilot gaze object recognition method and system based on eye movement data. The method comprises the following steps: sampling video data into images, constructing a training data set, training a YOLOV8n neural network, and obtaining a YOLO lightweight target detection model; based on the YOLO lightweight target detection model, frame-by-frame recognition is performed on the video data to be recognized; according to the relative position relationship between the gaze objects, the preliminary recognition result is checked and corrected; the pointer position of the instrument type gaze object is detected, and the gaze area of the gaze object is reconstructed according to the pointer position; and according to the gaze point coordinates in the table data and the reconstructed gaze area, the gaze object at each moment is determined. According to the table data and the video data recorded by the eye tracker, the pilot's gaze object at each moment in the flight training process can be accurately recognized.

The invention relates to the technical field of photovoltaic panel detection, in particular to a photovoltaic panel defect detection method based on airborne thermal infrared images. The method comprises the following steps: acquiring multi-view thermal infrared images collected by the unmanned aerial vehicle in at least two different flight attitudes, and generating a normalized temperature difference field for eliminating environmental and geometric interferences through registration and processing; segmenting and determining a thermal anomaly region based on the statistical characteristics; internal defects and surface covering defects are automatically distinguished by calculating angle invariance indexes of the thermal behavior characteristics under multiple view angles; respectively matching an internal defect feature library or a surface covering feature library; the detection reliability is evaluated by calculating the confidence coefficient of decision, and when the confidence coefficient is insufficient, the reason is analyzed according to the infrared image resolution and the feature variation, and the flight height, the hovering duration or the aerial photography angle of the unmanned aerial vehicle is adaptively optimized to recollect data. According to the method, the problem that real faults and pseudo defects in the photovoltaic panel cannot be automatically and reliably distinguished is solved.

The application provides an automobile door cover abnormality processing method and device. The method obtains a target door cover image and performs feature extraction to obtain target image features. A target framing region is determined from the target door cover image based on a door cover type. The target framing region is regionally segmented to obtain a plurality of initial segmentation regions. The defect types of the initial segmentation regions are determined according to the region colors and region shapes corresponding to the initial segmentation regions. If there are multiple defect types in any initial segmentation region, the initial segmentation region is continuously segmented until there is one defect type in each initial segmentation region, and a target segmentation region is obtained. The target region positions of the target segmentation regions corresponding to each defect type are counted. The target segmentation regions are processed based on the target region positions corresponding to each defect type and a defect processing method, so as to complete the abnormality processing of the target automobile door cover. The above method provides an automatic and efficient automobile door cover abnormality processing method.

The invention discloses a household fault intelligent identification method and system, and relates to the technical field of household fault intelligent identification, and the method comprises the steps: collecting the power utilization data of a user, and carrying out the preprocessing; analyzing the preprocessed data by using an anomaly detection algorithm, and identifying an abnormal condition; extracting the preprocessed data through a deep learning algorithm to generate a high-dimensional feature vector; and when an abnormal condition is identified, early warning is carried out, the high-dimensional feature vector is analyzed, and a fault mode is identified. According to the method, through comprehensive calculation of various factors such as the path length and the feature weight, the combination condition of various abnormal features is accurately recognized, the anomaly detection precision is remarkably improved, and the problems of missing detection and false detection caused by simple threshold judgment in the prior art are solved.

This invention discloses a method and system for detecting life jacket wearing at port berths. The method includes: constructing a monitoring area and deploying computing nodes; training a YOLOv8 model for initial and secondary detection, a ResNet model for illumination classification, a Uformer model for rain and fog recovery, and a YOLOv8 model for semantic segmentation; performing illumination scene recognition on the acquired images and adaptively selecting the corresponding preprocessing method; using the initial detection results, for personnel targets with insufficient pixel size, predicting their position by tracking and calculating the magnification factor, controlling the PTZ camera to magnify and perform secondary detection for confirmation; using the semantic segmentation results to determine whether the personnel are located in the land-based work area; and combining target tracking, triggering an alarm only when the personnel are not wearing life jackets, are located on land, and their status continuously exceeds a threshold. The system includes an intelligent PTZ camera, a management platform, and a processing unit. This invention solves the problems of difficult identification of small targets at long distances, poor adaptability to complex illumination, and high false alarm rate due to rule-based methods.

The utility model relates to the technical field of cloth detection, in particular to a cloth spreading mechanism for a cloth inspecting machine, which comprises a rack, a flattening mechanism is mounted at the top of the rack, a base plate mounted on the inner side of the rack is arranged below the flattening mechanism, and a tensioning component is movably connected in the rack; the flattening mechanism comprises a support fixedly connected to the top of the rack, a motor is installed at the top of the support, a driving shaft is fixedly connected to an output shaft of the motor, and a flattening assembly and a flattening assembly are connected to the driving shaft; according to the cloth spreading mechanism for the cloth inspecting machine, the motor is used for driving and enabling the smoothing assembly and the flattening assembly to operate, the smoothing assembly reciprocates on the surface of cloth through a smoothing plate, so that the cloth is flattened, then after the cloth passes through the flattening assembly, due to the fact that two sets of auger blades are symmetrically arranged on a flattening roller, the auger blades provide pulling force towards the two sides for the cloth, and the cloth is flattened; therefore, the cloth can be further flattened, and the problem of leak detection caused by wrinkles or partial overlapping on the cloth when the cloth inspecting machine works can be solved.

The invention provides a tunnel harmful gas scanning type monitoring method and equipment, and relates to the technical field of tunnel construction safety monitoring. Comprising the following steps: S1, synchronously scanning by at least two monitoring devices, recording direction angle data and a gas concentration measurement value of each ray, and acquiring tunnel boundary point cloud data; s2, calculating potential intersection points of rays from different devices in the tunnel space, and screening effective intersection points; s3, based on the tunnel boundary point cloud data, constructing a three-dimensional network model of the tunnel surface through a Delaunay triangulation algorithm; s4, in the three-dimensional grid, performing grading treatment according to concentration values from low to high, assigning values to voxels through which effective laser rays penetrate, and reconstructing a three-dimensional concentration field; and S5, generating a tunnel harmful gas concentration distribution three-dimensional model. According to the method, the path integral concentration data is converted into accurate three-dimensional space distribution information, the three-dimensional coordinates of the gas escape source can be accurately identified, accurate positioning of the harmful gas escape source is realized, and the problems of missing detection and misjudgment caused by fuzzy positioning are effectively solved.

The invention relates to the technical field of remote sensing image road extraction, in particular to a remote sensing image road extraction method based on frequency domain feature extraction and fusion, which comprises a feature extractor, a bottleneck layer and a segmentation branch. The feature extractor adopts FCB to replace traditional convolution operation, and performs frequency domain filtering operation on a frequency domain to perform feature extraction; the bottleneck layer adopts a KAN Layer to learn complex road features; according to the segmentation branch, the feature map is recovered to the original size through up-sampling, multi-scale feature learning and fusion are achieved through a multi-scale pyramid, and finally the feature map is sent to an output layer to output a prediction result. According to the method, the frequency domain convolution with the global receptive field, the bottleneck layer capable of learning the activation function and the multi-scale image reasoning fusion mechanism are introduced, the extraction precision is improved, the model parameter quantity and the energy consumption are reduced, and the high-efficiency and high-precision remote sensing image road real-time extraction requirement can be met.

The utility model relates to the technical field of textile, and particularly discloses a hairiness detector which comprises a back plate, a shell, an emitter, a receiver, an aviation plug, two first bolts, an installation assembly, a yarn guide assembly and a guide assembly, the shell is arranged on one side of the back plate, the back plate is provided with two installation grooves, one ends of the two first bolts penetrate through the corresponding installation grooves respectively, and the other ends of the two first bolts penetrate through the yarn guide assembly. The installation assembly and the yarn guide assembly are respectively in threaded connection with the shell, the emitter is fixedly connected with the shell and located in the shell, the receiver is fixedly connected with the shell and located above the emitter, the back plate is provided with a groove, one end of the aviation plug is fixedly connected with the shell, the other end of the aviation plug penetrates through the groove, and the installation assembly and the yarn guide assembly are both detachably connected with the back plate. According to the weft yarn hairiness detection device, the weft yarn hairiness can be accurately detected, the false alarm rate and the omission ratio are greatly reduced, and therefore the cloth quality is improved.

The utility model provides a kind of optical detection device for mobile phone middle frame appearance flaw, including outer frame, the inside fixed mounting of outer frame has arc hanging plate, the surface of arc hanging plate is installed with multiple first industrial camera and first detection light source, the middle part of one side of arc hanging plate is installed with extension plate, the surface of extension plate is installed with second industrial camera, the inside wall of the front side of outer frame is installed with multiple second detection light source and third detection light source, the bottom of the front side of outer frame is installed with third industrial camera, first detection light source, second detection light source and third detection light source constitute multi-angle light source illumination, imitate human eye, adapt to mobile phone middle frame appearance flaw detection, solve the problem of single light source angle in traditional appearance detection, to solve the quality anomaly and missed detection caused by part of defect not being imaged due to single angle.