93results about How to "Solve the problem of false detection" patented technology

The invention discloses a printing image defect detection method, which comprises the following steps: carrying out real-time image model learning in a gray scale region and a gradient region by aiming at specific images on a large image; comparing the large image to be detected to the a gray scale region model and a gradient region model established during learning; realizing small-dimension strong-contrast defect detection through Blob cluster analysis; when no small-dimension strong-contrast defect is detected, dividing the large image into sub regions, and respectively calculating image integrated features; adopting a variable threshold method for carrying out threshold division on each sub region; carrying out Blob cluster analysis on divided images; and realizing the large-area weak-contrast defect detection. Compared with the prior art, the invention does not rely on a reference template, is not sensitive on the real-time imaging brightness change, can overcome the defects of missing detection, error detection and poor self adaptation of the template by using a reference template comparison detection method, and can simultaneously solve the problems of eliminating tiny wrinkles and blackspots in printing and papermaking industries.

The invention discloses a deep learning saliency detection method based on the global a priori and local context. The method includes the steps of firstly, performing superpixel segmentation for a color image and a depth image, obtaining a global a priori feature map of each superpixel based on middle-level features such as compactness, uniqueness and background of each superpixel, and further obtaining a global a priori saliency map through a deep learning model; then, combining the global a priori saliency map and the local context information in the color image and the depth image, and obtaining an initial saliency map through the deep learning model; and finally, optimizing the initial saliency map based on spatial consistency and appearance similarity to obtain a final saliency map. The method of the invention can be used for solving the problem that a traditional saliency detection method cannot effectively detect a salient object in a complex background image and also for solving the problem that a conventional saliency detection method based on deep learning leads to false detection due to the presence of noise in the extracted high-level features.

The invention relates to a method for improving the accuracy of a geomagnetic parking stall detector, which comprises the following steps: (1) acquiring a temperature characteristic curve of a geomagnetic sensor; (2) recording an initialized temperature value and data of the geomagnetic sensor when initializing the geomagnetic parking stall detector; (3) reading data of a current temperature sensor and the geomagnetic parking stall detector in real time; (4) judging whether the ambient temperature difference of a current temperature value and the initialized temperature value exceeds a temperature threshold or not and executing the step (6) if the ambient temperature does not exceed the temperature threshold; (5) carrying out temperature compensation on the data of the current geomagnetic sensor according to the temperature characteristic curve to obtain data of the geomagnetic sensor subjected to the temperature compensation; and (6) judging whether vehicles exist or not according to the data of the geomagnetic sensor. The invention solves the problem that mistaken detection on the geomagnetic sensor caused by the environment influence, unidirectional read of the geomagnetic sensor and read shake, improves the accuracy of the geomagnetic parking stall detection as well as the practicability and the reliability of the geomagnetic parking stall detector in an environment with larger temperature change.

The present invention discloses a pedestrian detection method based on semantic segmentation information, and relates to the field of a pedestrian detection method based on a neural network. The pedestrian detection method based on semantic segmentation information comprises the steps of: 1: inputting original RGB images in training set samples into a backbone network, inputting corresponding semantic segmentation images into branch networks, and setting a loss function of the whole network to complete training; 2: inputting the original RGB images in THE training set samples into the backbonenetwork which completes training to perform convolutional feature extraction and generate a multi-layer feature map; 3: inputting the multi-layer feature map into an area generation network which completes training to perform pedestrian candidate frame extraction and generate a pedestrian candidate area; and 4: performing classification and location of the pedestrian candidate area by a classification regression network which completes training to output detection result images including a pedestrian position surrounding frame. The problem is solved that pedestrians and backgrounds are difficult to be distinguished in a low-resolution condition to cause low detection precision in the current pedestrian detection, and the precision of pedestrian detection is improved in the low-resolutioncondition.

The invention discloses a multi-angle human face detecting method based on weighting of deformable components. The method includes a model building process and a human face detecting process. Different weights are set on different components of a human face according to different contributions of the different components to a detection effect, and effects on important components can be sufficiently utilized assuredly. The problem in terms of large-angle human face detection is resolved by a multi-model fusion method, omission ratio is decreased by means of reducing a threshold value in a detection process, and false detecting ratio is decreased by means of a skin color confirming mode in a model result fusion process. The multi-angle human face detecting method effectively resolves problems of omission and false detection in the human face detecting process based on the components, and can be widely applied to the fields of safety detection, identity authentication, intelligent transportation and the like.

The invention relates to a box camera and ball head camera linkage-based road side parking detection system. A minimal detection unit of the system consists of a box camera and a ball head camera of which both have intelligent video analysis capability; the minimal detection unit is formed either via integration or combination of the box camera and the ball head camera; combination of the box camera and the ball head camera requires installation of an exchanger device which enables the two to access a network at the same time; the whole system consists of N minimal detection units; rods that are L meters in height are erected every X meters on a sidewalk of a side face of a roadside parking area; one minimal detection unit is installed on each rod. According to the system and the method, composite application of a plurality of intelligent video detection technologies such as vehicle movement detecting and tracking technologies, vehicle license plate recognition technologies, personneldetection technologies and the like is adopted; road side parking scenes, vehicle license plate information, vehicle characteristic information, conductor and driver information and parking timing information are bound together; timing and charging accuracy and reliability and charging basis validity can be improved; problems of non-transparent timing and charging, payment refusal and the like ina conventional solution can be effective solved.

The invention discloses a checkerboard angular point automatic screening method of corner detection, which employs a camera to shoot checkerboard images, obtains an angular point set to be identified through routine angular point detection, furthermore searches for a real angular point subset, and finally outputs checkerboard real angular points. The method comprises: designing a rectangular formwork; determining whether an angular point is a checkerboard angular point through calculating the similarity between the shape formed by angular points and the rectangular shape; employing a dioid graph method to describe the position relation among the angular points; and utilizing graph theory knowledge to respectively create distance sub graph and direction sub graph searching angular points. In detecting routine angular points, when a set angular point threshold level is lower and excessive angular point detection exists, the method can automatically determine a real checkerboard angular point subset from a detected angular point set without artificial intervention, remove false angular points, improve angular point detection accuracy, and lay a foundation for subsequent processing such as camera calibration, three-dimensional reconstruction, etc.

The invention discloses a FAST-AIC-algorithm-based automatic first-arrival collecting method for micro-seismic signals. The method includes: providing a new identification method of micro-seismic events according to the fact that the propagation path and the travel time information of hydraulic fracturing perforation signals are similar to those of the micro-seismic signals; using a hydraulic fracturing wave detector arrangement manner and a single-channel signal local extreme value to design a precise time window near a micro-seismic triggering moment; using Curvelet conversion to filter data in the precise time window; performing first-arrival collecting in the precise time window through a FAST-AIC algorithm, wherein the FAST-AIC algorithm is obtained by optimizing and converting a traditional AIC algorithm, the linear combination manner of the arithmetic sum and the quadratic sum of a discrete real number sequence is obtained, repeated calculation is reduced, and calculation efficiency is increased greatly. The new method is used to perform first-arrival collecting on analog data and actual hydraulic fracturing micro-seismic monitoring data to verify the feasibility and correctness of the new method. Compared with the first-arrival collecting result of a conventional micro-seismic method, the new method has evident advantages.

The invention discloses an intelligent test system and method for a metering device circuit board, belongs to the circuit board detection technology field and solves a problem of low efficiency caused by manual detection existing in a metering device circuit board in the prior art. The intelligent test system comprises a programming device used for realizing programming of function programs into a to-be-tested circuit board, a display performance test device used for testing display performance of the to-be-tested circuit board, a function test device used for testing functions of the to-be-tested circuit board, an identification device used for identifying the to-be-tested circuit board according to test results of the display performance test device and the function test device, a circuit board conveying mechanism which is connected with the programming device and is used for conveying the to-be-tested circuit board to the programming device, and a main control unit used for controlling operation of the programming device, the display performance test device, the function test device, the identification device and the circuit board conveying mechanism. The intelligent test system is advantaged in that the intelligent test system is applicable to circuit board tests of various types of intelligent gas meters.

The invention relates to a testing method of a machine vision based micro-sound film concentricity online testing system. The machine vision based micro-sound film concentricity online testing system comprises an optical imaging subsystem, a digital image processing subsystem, a motion control subsystem, a sound film eliminating subsystem, an image capture card and a motion control card, wherein signals of the optical imaging subsystem are sent to the image capture card through an interface, and the image capture card captures a frame of sound film gray scale image data and sends the data to the digital image processing subsystem to be processed and judged; a signal output end of the digital image processing subsystem is connected with a signal input end of the motion control card, and a signal port of the motion control card is respectively connected with the motion control subsystem and the sound film eliminating subsystem; and the motion control subsystem and the sound film eliminating subsystem are intensively controlled by the motion control card. The machine vision based micro-sound film concentricity online testing system realizes the precise positioning of sound films and the elimination of unqualified sound films by the motion control card controlling a stepper motor, a servo motor, a sensor, an eliminating device and the like, thereby realizing the high-precision online real-time testing of the sound films.

The invention discloses a method and device for detecting a target object, and relates to the technical field of computers. A specific embodiment of the method comprises the following steps: extracting a sample picture set from a historical video file recorded in a to-be-detected area; performing target object detection on the sample picture set based on a first confidence coefficient threshold value and a second confidence coefficient threshold value which are different in size to obtain sample data sets under different confidence coefficient thresholds; determining an invalid region by comparing the sample data set under the first confidence threshold value with the sample data set under the second confidence threshold value; and when the currently acquired image data is detected, excluding the target object in the invalid area. According to the embodiment, the detection accuracy of the target object can be improved.

The invention discloses a method and a device for judging an error detection of a physical downlink control channel. The method comprises the following steps: making a hard decision on a soft-bit data output after a rate de-matching; decoding the soft-bit data and carrying out cyclic redundancy check, coding the decoded data after the cyclic redundancy check; comparing the coded data with the result of the hard decision, calculating the ratios of different bits between the coded data and the result of the hard decision to the soft-bit data; judging whether the ratio is greater than a scheduled threshold value or not, if so, judging to be error detection. According to the technical scheme, the problem that PDCCH (physical downlink control channel) error detection can not occur under an LTE (long term evolution) system can be solved, and the probability of the occurrence of the PDCCH error detection is greatly reduced.

The invention discloses an optical satellite remote sensing image cloud detection method. The method comprises the steps of counting a remote sensing image gray histogram and preprocessing the remote sensing image gray histogram; calculating a Gaussian mixture model component and an iteration maximum between-class variance threshold value according to the gray histogram, and further calculating a mixture threshold value; segmenting the remote sensing image by using the mixed threshold to obtain a cloud initial detection image; obtaining a contour boundary characteristic value and a cloud shadow matching characteristic value of a cloud area in the cloud initial inspection image, further removing the cloud area which is mistakenly detected, and obtaining a mistakenly overhauled positive image; performing missing detection thin cloud compensation on the cloud initial detection image to obtain a thin cloud compensation image; and carrying out comprehensive processing on the false overhaul positive image and the thin cloud compensation image to obtain a final cloud detection image. According to the method, the cloud area can be rapidly and accurately detected without manual or other auxiliary data participation, the cloud area which is mistakenly detected is removed in the detection process, the thin cloud area which is missed is compensated for, the cloud detection precision is improved, and the method is suitable for panchromatic remote sensing images and multispectral remote sensing images.

The invention disclose a method for detecting a vehicle queue length based on a local variance, which is applied to the field of intelligent transportation, and particularly relates to a digital image processing and computer visual technique. The method comprises the following steps: putting forward an improved local variance method for performing vehicle detection; putting forward a combination of the improved local variance method and an LBP (Length between Perpendiculars) algorithm to remove vehicle shadows; and putting forward a method for reducing vehicle queue length calculation errors. The method for detecting the vehicle queue length has a small calculation quantity, and the problem of error detection of vehicles caused by vehicle shadows of lanes during multi-lane detection is solved. Due to the adoption of the method, functions of real-time control of traffic lights, measurement of traffic flow control indexes and the like can be realized; and the method plays important roles in maintaining traffic safety and urban security, increasing the road utilization ratios, reducing traffic jams, realizing automatic traffic management and intelligent city, and the like.

The invention relates to an atrial fibrillation detection method and system. A condition that whether signals have P wave or not is added while the characteristic reflecting RR interval irregularity of the ECG signals is extracted, so that the problem of false detection of sinus arrhythmia signals is solved; a BP neural network method is adopted to replace simple logic conditions to judge atrial fibrillation signals, so that atrial fibrillation detecting accuracy when input RR interval number is small is improved.

The invention discloses a medicine box detection method based on three-dimensional point cloud and image data fusion, and the method comprises the steps: inputting a medicine box image collected by a camera into an optimized U-shaped full convolutional neural network, and extracting a medicine box feature image; in the feature extraction part, a grouping residual error convolution module is used for extracting a preliminary feature image, a cavity space convolution pooling pyramid module is used for extracting feature image information of different scales of the preliminary feature image, and a mixed attention module is used for fusing the feature image information of different scales to obtain a two-dimensional fusion feature image; a segmented medicine box image is obtained through up-sampling; and whether the segmented medicine box image meets the detection requirement is judged, if not, three-dimensional information of the medicine box is extracted, a target in the image is positioned through a two-dimensional target detection network, and a cone point cloud corresponding to the two-dimensional detection frame is obtained according to a camera geometric imaging model. A Point Net point cloud network and a feature fusion network layer are adopted to carry out instance segmentation on the cone point cloud to obtain all target points. A target centroid is estimated by using a T-Net network, a target point cloud is moved to a centroid coordinate system, then estimation of parameters of a three-dimensional bounding box is obtained through a parameter estimation network and a feature fusion network layer, finally, the size and orientation of a medicine box are obtained, the type of the medicine box is judged, and the character 0 of the medicine box is recognized in combination with image information. The problems of long time consumption, high false detection rate and the like in existing medicine box detection are solved.

The invention relates to an image tampering region detection method based on secondary JPEG compression. A texture smooth region mask is introduced in order to solve the problem of the easy false detection of a texture smooth region of an image. By use of the mask and a JPEG error, a tampering probability graph is generated, so the problem of the false detection of the texture smooth region is solved. Meanwhile, in order to solve the problem of inaccurate estimation of a compression quality factor of a low-quality image, in the image tampering region detection method, a JPEG Ghosts characteristic obtained through recompression of the compression quality factor of the low-quality image is not directly used, but JPEG Ghosts characteristics under multiple compression quality factors are converted into tampering probability graphs and multiple remarkable tampering probability graphs are selected and combined into a tampering probability graph. According to the image tampering region detection method, the robustness is strong, false detection is less and computing speed is fast.

The invention relates to a method for controlling a desaturation detection protective circuit in a T-shaped three-level inverter. The method comprises the steps of: analyzing and judging an output voltage current signal of the inverter through sampling, filtering, amplifying and comparing circuits by a digital signal processor to determine the direction of the output voltage current; sorting the output state of voltage current combination in different directions: area I (v)0,i)0), area II (v)0,i(0), area III (v(0,i(0) and area IV (v(0,i)0), respectively; then judging the positive and negative semi-cycle reversing condition in each state of current, including positive reversing to negative reversing and negative reversing to positive reversing; carrying out enabled logic control on the desaturation detection protective circuit of a switch under different output states and current reversing conditions; and receiving an enabled control signal by a switching drive chip or a peripheral circuit, so that a desaturation detection function is alternately activated and inhibited when the inverter converts and inverses. The method avoids a spurious triggering fault of the desaturation detection circuit caused by factors such as oscillation in the converting and reversing process of the T-shaped three-level inverter, and the reliability of fault diagnosis by the system is effectively improved, so that the normal work of the circuit is guaranteed.

The invention discloses a high-speed rail overhead contact line equipment object detection and tracking method based on Kalman filtering. The method comprises the following steps: acquiring sample images; preprocessing the sample images; for the preprocessed sample images, obtaining pixel points on a horizontal scanning line and performing saliency ordering, and extracting salient pixel points; based on the extracted salient pixel points, detecting catenary wires and lead wires in the preprocessed sample images; determining a catenary wire and lead wire cooperative tracking strategy, and tracking the catenary wires and the lead wires; and based on the Kalman filtering, predicting an object movement locus, and correcting a tracking result. According to the invention, based on a Kalman filtering principle, through detecting and tracking the catenary wires and the lead wires in a complex background, overhead contact line equipment is monitored in real time, early warnings can be effectively given in case of emergencies in a driving process such as pantograph-catenary disconnection and the like, and accidents caused by pantograph-catenary slip are avoided to a certain degree.

The invention provides a pedestrian abnormal behavior detection method based on a human skeleton, and the method comprises the following steps: 1, carrying out the pedestrian detection of an image, carrying out the framing through a detection frame, and obtaining a pedestrian detection frame; 2, extracting human skeleton information from the pedestrian detection box obtained in the step 1 to obtain a human skeleton information picture, and performing background removal preprocessing on the picture; and 3, performing pedestrian abnormal behavior detection on the preprocessed human body skeletoninformation picture in the step 2 by using a multi-scale information fusion recognition network based on residual errors to obtain a four-dimensional vector corresponding to four types of actions ofhuman body abnormal behaviors; according to the method, the skeleton information and the multi-scale information are effectively utilized, the detection accuracy of the algorithm is greatly improved,the robustness of the algorithm is enhanced, and the application of the algorithm is not limited by scenes.

The invention provides a method and device for detecting a filament ingot and a filament ingot sorting system. The method comprises the steps: obtaining a to-be-detected filament ingot image, inputting the to-be-detected filament ingot image into a target detection model employing an improved YOLO network based on depth separable convolution and point convolution, and carrying out the detection ofa defect region and / or an interferent region; when it is determined that the defect area and / or the interferent area are / is detected, specified reference data of the defect area and / or the interferent area are / is extracted; determining whether the filament ingot in the filament ingot image is a defective product or not according to the specified reference data and preset standard reference data corresponding to the defect sample area; and when it is determined that the defective products are defective products, the defective products are sorted to a defective product area, defect informationof the defective products is output and displayed, and the defect information comprises position information of the defective area. According to the method and device for detecting the wire ingot andthe wire ingot sorting system, the problems of missing detection and false detection existing in an existing method for detecting the surface defects of the wire ingot are solved, and the problems that the detection cost is high and the detection efficiency is low are solved.

The invention provides an injection molding mechanical arm mold anomaly detection method based on an LMDO (Local Multilayered Difference Operator). The anomaly detection method comprises the following steps: (1) acquiring a standard template image when an injection molding machine opens a mold in place, and pre-processing to obtain a later difference background image; (2) waiting for working state information of the injection molding machine; upon detection of a situation that the injection molding machine is operated until the mold is opened in place, continuously acquiring the image of a mold cavity by a camera, extracting an average image of the plurality of images, and pre-processing the average image to do preparation for subsequent image processing, thereby obtaining a later difference foreground image; and (3) carrying out an anomaly detection algorithm based on the LMDO on the difference foreground image and the difference background image to obtain an abnormal region without a light illumination interference part. The injection molding mechanical arm mold anomaly detection method based on the LMDO, provided by the invention, has the characteristics of good instantaneity, strong robustness on illumination variation and the like; and whether the mold has an abnormal state or not can be monitored through mold opening information of the injection molding machine.

The invention provides a false detection preventive method for downlink control information in an LTE (-A) system. The method comprises that PDCCH is decoded, CRC verification is carried out after decoding, DCI is determined to be incorrect if CRC verification fails, and whether false detection occurs in the DCI is determined if CRC verification passes; DCI decoding information and DCI format information corresponding to the DCI decoding information in the PDCCH decoding process are cached, the DCI decoding information is determined by utilizing a false alarm determination rule corresponding to the DCI format information, DCI is determined to be incorrect if the false alarm determination rule is not met, and DCI false detection prevention determination is carried out if the false alarm determination rule is met; and the average power of DCI is obtained, the adjacent average power of CRS is obtained, the average power of the CRS is compared with the average power of the DCI, the DCI is determined to be incorrect if the difference therebetween is greater than a preset threshold, and the DCI is determined to be correct if the difference therebetween is lower than the preset threshold.

The invention provides an image acquisition device-based vending machine, which comprises the following steps of (S11) confirming user registration information through recognition, obtaining a picturein the vending machine before a door is opened through photographing of a camera, uploading the picture to a processing server as an original image; (S12) confirming the situation of articles beforethe door is opened through the processing server and transmitting a door opening signal; (S13) confirming that shopping of a user is completed through receiving a door closing signal, obtaining another picture in the vending machine after the door is opened through photographing of the camera and uploading the picture to the processing server as a current image; (S14) confirming the situation of the articles after the door is closed through the processing server, and comparing the situation of the articles before the door is opened with the situation of the articles after the door is closed toobtain the sold articles; and (S15) calculating and displaying to-be-paid amount through price information of the sold articles and cutting payment from an account of the user registration information. The image acquisition device-based vending machine has the advantages that ''grab-and-go and flexible pickup'' are achieved and the image acquisition device-based vending machine is convenient to use and accurate in checkout.

The invention discloses a cloud mask-oriented non-uniform category sample equalization method. The method comprises the steps of 1, data preprocessing and sample acquisition; 2, sample grouping; 3, cloud mask model training; 4, classifier masking and evaluating; 5, iterative training; 6, cloud masking; and 7, mask data post-processing. According to the sample equalization method adopted by the invention, sample imbalance caused by uneven cloud categories in the remote sensing image is effectively solved, and effective identification and segmentation of various types of clouds in the image arerealized, so that the cloud mask precision is improved; the samples are selectively input, so that the influence of missing or wrong detection samples can be highlighted, the small-class samples are enhanced, the features extracted by the deep learning model are effectively adjusted, and the problem of missing detection or false detection of small-class clouds is solved.

The invention discloses an eddy current detection uplift shake compensation method. An eddy current detector employs time-sharing switching of a main check excitation signal and an auxiliary excitation signal to excite an eddy current detection coil, the auxiliary excitation signal monitors the uplift or shake change of the eddy current detection coil, through a set effective detection uplift shake signal threshold, missing detection or false detection caused by eddy current detection coil uplift or shake change can be effectively solved, and further, the auxiliary excitation signal monitors the uplift change of the eddy current detection coil, so that defect signals can be corrected and compensated in real time.

The invention discloses a method for extracting and identifying the end surface defects of a steel pipe, which comprises the following steps: (1) adopting a reasonable light source construction mode of a camera, firstly dividing a binary diagram of a chamfered part of the steel pipe by a fixed threshold value; (2) fitting the chamfered inner and outer diameters of the end face of the steel pipe bythe least square method to determine whether the chamfered angle of the end face of the steel pipe is reversed or not; (3) creating a mask image according to the chamfered inner and outer diameter parameters fitting in step 2; (4) performing AND operation on the mask image and the original image obtained in the step (3) to obtain an image containing only the end plane and no chamfer; (5) using the OTSU algorithm to segment the end plane image according to the image obtained in the step 4 to obtain the binary image of the defect region. (6) By calculating the characteristic parameters of the defect region and using the support vector machine, the defect types can be identified. The invention has the advantages of high operation efficiency, strong stability and wide probationary performance, and the method can be used for the end surface defects of most steel pipe types or ring products.

The invention provides a saliency detection method based on depth selective difference. The method comprises the steps: depth images are acquired as input, smooth process is performed on each depth image, a selective difference value of each segmentation area are calculated, and an initial saliency image is optimized according to center preference, so that a final saliency detection result is obtained. Use of the method solves a problem that an object of which visual features are similar with the background can not be detected purely based on a color image and a problem that a bottom background area ignored based on a depth image causes false detection. According to the invention, the method is suitable for saliency detection method of the depth image, the computation complexity is relatively low, the detection result is accurate, and the method has wide application in the field of image processing and computer vision.

The invention discloses a one-in-multiple circuit and a one-in-multiple method of a clock and belongs to the technical field of clock selection in a spaceflight electronic system. The clock one-in-multiple circuit is composed of a clock counter module, a clock selection signal generator module and a clock selector module, wherein the clock counter module, the clock selection signal generator module and the clock selector module are connected in series. A field programmable logic gate array is utilized to autonomously and reliably screen shared-frequency multi-clock sources. The clock counter module performs cycle count on each way of input clock signals, and the clock selection signal generator module detects the validity of the clock signals and outputs a clock signal selection signal. A clock selector module chooses the input clock signals and outputs one way of the clock signals. According to the one-in-multiple circuit and the one-in-multiple method of the clock, the triplication redundancy technology is utilized by a register in the circuit, the reliability of the circuit when the single event upset effect happens is enhanced, the problem of wrong detection when multi-to-one detection is performed on shared-frequency clocks in the prior art is solved, system cost, complexity and the degree of coupling are reduced, and the system efficiency and reliability are improved.

The invention discloses an electromagnetic adsorption type ultrasonic flaw detection robot, and relates to the technical field of intelligent ultrasonic flaw detection equipment. The robot comprises acarrying rack, a crawling mechanism and a pressing mechanism, the crawling mechanism comprises a plurality of mechanical legs hinged to the left side and the right side of the carrying rack, and thepressing mechanism comprises a plurality of vertically-arranged pressing modules fixedly connected to the interior of the carrying rack; each pressing module comprises rectangular grooves fixedly connected with the two ends of the pressing module, probe support fixing plates clamped in the rectangular grooves, probe supports fixedly connected with the lower surfaces of the probe support fixing plates, probes fixedly connected with the lower surfaces of the probe supports, and probe pushing devices controlling the probe support fixing plates to move up and down. Hinged shafts at the hinged positions of the mechanical legs and the carrying rack are vertically arranged. The operation process is simple and efficient, the stability of equipment is improved, and the problems of low manual flaw detection efficiency, high labor intensity and the like are effectively solved.