1296 results about "Image edge" patented technology

The invention provides an infrared and visible light image fusion method based on multi-scale feature interaction enhancement, belongs to the field of image processing, and aims to improve the image fusion quality in a complex environment. The method comprises the following steps: firstly, performing feature extraction on infrared and visible light images through a double-flow multi-scale encoder to generate multi-level features of different scales; and then, cross-modal information interaction is carried out by adopting a feature interaction enhancement module, fusion features are optimized through a space and channel attention mechanism, and clear expression of key areas such as target heat radiation and image edges is ensured. In the decoding stage, multi-scale jump connection and layer-by-layer up-sampling are utilized to reconstruct a high-resolution fusion image. And finally, training the fusion network through a content-gradient joint loss function to realize balance between detail reservation and global information. The method is suitable for tasks such as monitoring and target identification in a complex scene, and can effectively improve the image fusion effect especially under the conditions of low illumination, smoke shielding and the like.

The invention relates to the technical field of image processing, and discloses a backlight effect image edge enhancement method based on intelligent identification, which comprises the following steps of: judging a field environment type; performing global optimization on the original image based on a set environment perception type enhancement mechanism according to the judged field environment type, and outputting a global pre-processing image; constructing a backlight area segmentation model for the globally preprocessed image; outputting a local enhanced image; designing a structure perception type local adaptive threshold algorithm for the local enhanced image, and outputting a binary image keeping structural continuity; extracting an edge image of the binarized image through an edge detection algorithm, and optimizing a topological structure of a contour in the binarized image; and comparing with a wood template processing standard feature library, outputting an edge quality evaluation result and feeding back to a processing control system. Defect detection and machining control depth linkage is achieved, passive detection is changed into active optimization, and the production efficiency and the yield are improved.

The invention relates to the technical field of image defect detection, in particular to a building engineering quality detection and evaluation system based on machine vision, which comprises an image acquisition module, a crack main shaft identification module, a space packaging module, a structure overlapping detection module and a building quality evaluation module. According to the method, a crack boundary layer with a spatial positioning function is constructed through image edge scanning and pixel position marking, pixel segments with high continuity are screened to generate a path line, the continuity and representativeness of crack trend are improved, and a rectangular block is packaged to enhance organization and complete expression of cracks in a spatial level; performing coordinate comparison with a structure stress area to form corresponding association between defect influence and a structure function, judging the influence of the crack on a key area through a space coverage relation between a total block amount and an overlapping amount, and establishing multi-dimensional building quality grade division by combining crack form, position distribution and structure association information; and comprehensive evaluation of building structure defects from geometry to structure level is realized.

The invention relates to the technical field of nondestructive testing, in particular to a leak detection and partial discharge digital imaging detection method and system based on acousto-optic fusion, and the method comprises the following steps: based on channel microphone array sound wave data in a partial discharge signal suspicious region, extracting a sound wave abnormal section, positioning a sound source, matching image edge features, and synchronously marking; and analyzing the phase change of the multi-frequency signal to judge a sound source concentration area, tracking the moving trend of a disturbance point, and outputting an acousto-optic fusion positioning trend track. According to the method, the partial discharge feature recognition sensitivity is improved through high-frequency peak paragraph screening and dominant frequency recognition, the abnormal region positioning precision is enhanced in combination with image edge extraction and sound source space matching, and acousto-optic synchronous positioning and trend trajectory display are achieved through multi-frequency signal phase analysis and image frame disturbance tracking. Through fusion of frequency domain feature extraction, image recognition, dynamic comparison and other actions, the spatial precision of abnormal source recognition and the multi-source fusion analysis efficiency are improved, and the partial discharge traceability and dynamic monitoring capability are enhanced.

The invention relates to the technical field of power transmission line detection, and discloses a power transmission line key component defect identification method based on cloud edge cooperation. The method comprises the following steps: acquiring a multi-source inspection data set consisting of a power transmission line component image acquired by an unmanned aerial vehicle, edge end sensor data and a cloud historical defect database; at the edge end, extracting component region features through an image preprocessing algorithm, and generating environment correlation parameters by using a multi-modal feature fusion algorithm; and at the cloud, performing space-time correlation analysis on the historical defect database to generate a part defect evolution graph. And inputting the information into a cloud edge collaborative recognition model to obtain a component defect feature vector, constructing a multi-stage defect recognition network through a dynamic optimization algorithm, and outputting a component defect classification result and confidence. According to the method and the system, the accuracy, the real-time performance and the reliability of defect identification of the key component of the power transmission line are improved, and the method and the system have good application prospects.

The invention relates to the technical field of automatic driving, in particular to an automatic homeward voyage method for multi-sensor data fusion of an unmanned vehicle, which comprises the following steps: acquiring laser radar obstacle points, image edges and attitude data, completing synchronous registration, identifying space mapping, extracting track and boundary trends, screening forward sections, and judging track offset and trafficability. Screening an error band boundary, adjusting a steering instruction, analyzing path overlapping change, and outputting a path keeping label. According to the method, through time synchronization and space registration of various sensor data, high-precision fusion of the data under a unified space-time framework is realized, the reliability of environment perception is improved, accurate mapping of the data in a unified reference coordinate system is ensured, the reliability of path planning and decision making is enhanced, and the optimal advancing direction is determined by combining a homing point; the accuracy of return route planning is improved, the traffic feasibility is dynamically evaluated, the effective error band boundary is screened to improve the safety, and the reliability of the return route is ensured through track stability analysis.

The invention provides a rock image classification method based on edge enhancement and multi-scale feature fusion. By combining the edge enhancement and multi-scale feature fusion technology, the accuracy of mineral classification in the rock slice image is remarkably improved. According to the method, rock slice image interference is eliminated through filtering and denoising, and mineral particle edges are enhanced by fusing morphological top-hat transformation and a Laplace operator; and performing multi-scale pyramid decomposition on the enhanced image to extract high-frequency information, performing multi-direction response enhancement to generate a direction feature map, and performing channel-level fusion on the original image, the edge enhanced image and the direction feature map. And finally, inputting the fused image into a convolutional neural network to complete rock type classification. According to the method, mineral boundary expression is effectively enhanced, the classification accuracy is improved, and a reliable technical scheme is provided for geological analysis and lithology identification.

A method for measuring the size of a slide block of a guide rail with a complex shape based on intelligent vision belongs to the field of machine vision, and comprises the following steps: DXF file data acquisition: reading by using Python; image acquisition and preprocessing: acquiring image data in real time by using an industrial camera, and performing image preprocessing; hole detection and measurement: adopting a Hough transform algorithm to realize sliding block hole detection, introducing a sub-pixel counting method to improve the algorithm, and realizing aperture high-precision measurement; edge line detection and measurement: extracting edge and line segment features of the sliding block image by adopting Canny edge detection and LSD line segment detection technologies, automatically removing invalid line segments, determining an optimal fitting line segment, and realizing accurate detection and measurement of a target edge line; and detection and measurement result visualization: displaying a sliding block size measurement result in an image, comparing the sliding block size measurement result with a standard parameter of a DXF file, judging the qualification of the sliding block, constructing a visual interface by adopting Pyside2 to record the measurement result, and supporting a plurality of input source detection functions.

The invention relates to the field of phytoplankton image edge detection, in particular to a marine phytoplankton image edge detection method based on a non-local mean denoising mode. The method comprises the following steps: putting a marine phytoplankton image file needing to be detected into a folder of a program item, automatically reading a code into an image in the folder by a program, and loading a to-be-processed image; edge details related to the marine phytoplankton image are reserved based on a non-local mean denoising algorithm; aiming at the image data after the edge details are reserved; carrying out edge extraction on the image based on a Canny edge detection algorithm; and outputting a marine phytoplankton image with edge details reserved. Through combination of non-local mean denoising and Canny edge detection, noise can be more effectively removed, image details and structure information can be kept, and the influence of noise on edge detection is reduced, so that clearer and more accurate edges are extracted; and in an image with relatively large noise, the edge extraction precision and the image quality can be remarkably improved.

The invention provides an image quality enhancement method and system based on a cloud game, and the method comprises the following steps: firstly, a cloud end renders a game picture and generates an initial video stream; then, performing preliminary optimization on the video stream, and reducing jumping and frame loss through motion compensation of an inter-frame motion vector; the optimized video stream is transmitted to a user side, after the user side receives the video stream, image quality enhancement processing is carried out, specifically, denoising processing is carried out, and random noise and compression noise are eliminated by combining an inter-frame difference method and spatial domain bilateral filtering; sharpening enhancement: enhancing image edges and texture details by using a Laplacian operator; artifact removal: removing compression artifacts through frequency domain filtering; the dynamic contrast is enhanced, and gray level distribution is adjusted through histogram equalization; finally, the video stream after enhancement processing is presented to a user in real time, the image quality of the cloud game can be remarkably improved by gradually updating the processing result of each pixel point, the video stream is presented in real time in a high-quality mode, and the user experience is optimized.

The present invention relates to the field of computer vision technology, and specifically to a pathological slice image analysis method and system, comprising the following steps: based on pathological slice image information, combined with image preprocessing, edge information in the image is detected and extracted, color information is analyzed, color distribution characteristics in the image are evaluated, and image edge and color information is generated. In the present invention, by analyzing edge and color information, combined with the extraction of texture features, the identification process of tissue structure is optimized, the identification and segmentation of lesion tissue is made more accurate, the identification and segmentation of lesion tissue is performed using tissue structure feature data, the location of the lesion area and the identification of the type are achieved, and the visual presentation quality of the data and the intuitiveness of the diagnosis are improved by mapping the pathological slice image into a three-dimensional model and annotating it, helping medical professionals to fully understand the pathological conditions in the slice image and make accurate treatment decisions.

The invention discloses a medicine image detection method based on an improved Canny operator, and the method comprises the steps: 1) collecting a medicine image, and carrying out the graying processing of the medicine image; 2) an improved bilateral filter is adopted to filter the medicine image after graying processing, and noise is suppressed; 3) carrying out gradient amplitude and phase extraction on the medicine image by adopting a four-direction Soble operator; 4) performing non-maximum suppression on the extracted gradient amplitude and phase by adopting improved non-maximum suppression, and accurately positioning a target edge; 5) performing adaptive threshold extraction by using Otsu improved based on gradient edge information to obtain a drug image edge contour; and 6) outputting the edge contour of the medicine image. By adjusting the weight value of the edge pixel point identified by the bilateral filter to be zero, the interference to the central pixel point is effectively reduced, so that the target edge and detail features of the image are reserved to the greatest extent.

The invention discloses a circuit breaker image edge detection method fusing spatial constraint fuzzy clustering and lightweight network optimization, and the method carries out the local adaptive threshold calculation through combining spatial constraint FCM and Otsu algorithms, and optimizes the edge detection process of a Canny operator. According to the method, fuzzy classification is carried out on a circuit breaker image by adopting spatial constraint FCM to obtain a strong marginal probability graph; the circuit breaker image is subjected to block processing through local adaptive threshold calculation, a global threshold is generated through integration, and then the global threshold is input into a Canny operator for accurate edge extraction. In order to further improve the detection effect, a lightweight neural network PiDiNet is used to correct a Canny output image. According to the method, the edge detection precision of the circuit breaker image can be effectively improved, and the method is suitable for edge extraction tasks in high-noise and complex background environments.

The invention discloses a large-size grinding wheel concentricity error measurement method and system based on machine vision, and relates to the technical field of grinding wheel concentricity error measurement, and the method comprises the steps: calibrating an industrial CCD camera through PyCharm software, collecting a part of images of a grinding wheel, and carrying out the RGB three-channel extraction, gray conversion, filtering and binary segmentation processing of the collected images; secondly, a Canny edge operator is adopted to carry out coarse positioning on the edge of the image, and then a novel edge judgment criterion is established to construct an edge detection model with sub-pixel-level precision; and finally, carrying out feature fusion on sub-pixel contours of RGB channels, clustering extracted discrete point sets by using an improved DBSCAN algorithm, and then carrying out ring fitting by using improved RANSAC to realize accurate measurement of the inner and outer diameters and concentricity of the grinding wheel.

The invention relates to the technical field of mining area safety monitoring, in particular to a mining area settlement monitoring method and system based on image data, and the method comprises the steps: collecting a mine hole image in a mining area, and carrying out the edge detection of the mine hole image, and obtaining an edge image; high-definition images and to-be-enhanced images are screened out according to the definition degree of the edge images, and the definition degree reflects the pixel value and the position change degree of pixel points at the same position of two adjacent frames of edge images. According to the method, the supplementary coefficient of each pixel point is calculated by utilizing the characteristics between the high-definition image and the to-be-enhanced image, the to-be-enhanced image is enhanced, so that a complete mine hole edge can be obtained in the to-be-enhanced image, and finally, the time when the mine hole edge is abruptly changed is calculated. According to the difference value between the area of the mine hole edge at the sudden change moment and the area of the mine hole edge at the current moment and the change time, the settlement coefficient of the mine area can be accurately calculated so as to monitor the safety of the mine hole.

The invention discloses a graph neural network embedding method combining a heterogeneous graph and multi-modal data, a medium and equipment, and the method is characterized in that the method comprises the following steps: carrying out the feature initialization of nodes and edges in the heterogeneous graph, employing a text pre-training model, an image convolution network or an audio processing model to initialize the node features according to the node types, carrying out weight distribution and feature coding on the semantic edge, the image edge or the social edge according to the edge type; multi-modal data is mapped to the same potential space through a modal self-adaptive fusion mechanism, weights of different modals are dynamically adjusted, and the modal self-adaptive fusion mechanism comprises construction of a shared embedding space and a multi-head attention mechanism; performing weighted aggregation on node information based on heterogeneous edge types, updating node embedding through propagation of a multi-layer graph neural network, and defining different weighting functions according to the edge types in an aggregation process; and embedding and applying the final node to a downstream task, and optimizing the model through a task-related loss function.

The invention relates to the field of infrared image processing, in particular to an infrared blind pixel detection method based on temperature self-adaption and time-frequency domain synthesis, which comprises the following steps of: traversing an infrared image by using a sliding window to obtain gray values of pixel points around a current pixel point; carrying out image edge judgment by using a Sobel-Otsu algorithm based on working temperature dynamic adjustment, and reserving a non-image edge window to participate in blind pixel judgment; grouping pixel points in the window according to horizontal and vertical directions; calculating the alpha filtering value of the gray scale of each group of pixels; calculating the deviation between the gray value of each pixel in the window and the alpha filtering value of the corresponding row coordinate and column coordinate group; if the deviation is greater than an adaptive threshold, setting the blind pixel as a potential blind pixel, and constructing a potential blind pixel distribution matrix; a time domain accumulation method for pre-screening periodic noise based on FFT is adopted, periodic noise interference is eliminated through frequency domain analysis, and then high-probability blind elements are screened through time domain accumulation; and obtaining a final blind pixel detection result. The method can effectively cope with the working temperature change of the sensor and the noise interference in a complex environment, and remarkably improves the accuracy and anti-interference capability of blind pixel detection.

The invention relates to the field of remote sensing image processing and computer vision, and discloses a satellite image cascade matching method and system based on double-branch context awareness, and the method comprises the following steps: constructing a heterogeneous double-branch feature extraction network, extracting multi-scale detail features through a main feature network, and capturing global semantic information through a context coding network; fusing multi-scale main features and context features to construct a group-related cost body, and adaptively enhancing key region matching response in combination with a channel incentive mechanism; optimizing the cost body by adopting a cross-scale information transfer and multi-dimensional attention fusion strategy, and generating a multi-scale initial disparity map; image edge geometric features and semantic contexts are fused, and high-resolution parallax details are recovered through a multi-stage residual decoder. According to the method, the problem of matching fuzziness of a traditional method in weak texture, repeated structure and parallax abrupt change areas of a satellite image is solved, and the three-dimensional reconstruction precision and the edge detail integrity in a complex scene are remarkably improved.

The invention provides a marine radar image denoising method and device based on a multistage filtering algorithm, and the method comprises the steps: carrying out the preprocessing of a to-be-processed marine radar image, and obtaining a first image; performing edge gradient detection on the first image based on the gradient direction corresponding to each pixel point in the first image and the target identified in the first image, and performing noise detection on the first image based on the gray value corresponding to each pixel point in the first image after the edge gradient detection is passed; after the noise detection is passed, processing the first image by adopting bilateral filtering to obtain a second image; and image enhancement is carried out after image sharpening is carried out on the second image, and a denoised marine radar image is obtained. According to the method, multiple filtering technologies such as Gaussian filtering, median filtering and bilateral filtering are combined, multiple kinds of noise in the marine radar image are effectively removed, especially image edge and target information can be kept in a complex environment, and the target detection and navigation precision is remarkably improved.

The invention records a multi-scale attention and median enhancement semantic segmentation method, and the method comprises the following steps: S1, extracting the initial features of an original image, and forming the output features of an Inception block; s2, inputting the output features into a multi-scale feature enhancement module, and outputting multi-scale features; s3, inputting the multi-scale features into a median enhancement space channel attention module, and outputting an attention feature map; s4, generating a final fusion feature of the high-level feature and the low-level feature through the cross feature fusion block; s5, performing up-sampling on the attention feature map, adding the attention feature map with the final fusion feature, and performing up-sampling to obtain a fusion feature map; s6, performing up-sampling and splicing on the fused feature map, and outputting a spliced feature map; and S7, performing up-sampling and full-connection layer expansion on the spliced feature map to obtain a segmentation prediction map of the original image. According to the method, feature distribution information can be captured more comprehensively, and a clearer and more accurate original image edge segmentation result is obtained.

The invention provides a saliency target detection method and system based on edge detection and an attention mechanism, relates to the technical field of computer vision, and utilizes a target detection network to carry out saliency target detection. The target detection network emphasizes the selectivity and invariance of the features during detection of a salient edge and a salient region through parallel edge branches and salient detection branches; the edge branch performs interactive fusion on the low-level features with spatial structure details and the high-level features with rich semantic knowledge to obtain edge features; the saliency detection branch generates a region feature containing multi-scale key information through multi-scale attention, and improves the interior and boundary of the region feature through an edge guide learning strategy by using edge features as guidance to obtain a final saliency target region; according to the method, edge detection and an attention mechanism are combined, saliency target identification and positioning and image edge definition are better realized, and the accuracy of saliency target identification of the system is improved.

The invention discloses an image edge cutting method and device based on boundary detection, equipment and a storage medium, and relates to the technical field of image boundary detection, and the method comprises the steps: carrying out the preprocessing of an original image, and obtaining a target image; determining a scene type of the target image; based on an edge detection strategy corresponding to the scene type, performing edge detection on the target image to obtain an edge image; performing straight line detection on the edge image to obtain a plurality of candidate straight line segments; and based on a boundary positioning rule, determining a target boundary line segment from the plurality of candidate straight line segments, so as to carry out edge cutting on the original image according to the target boundary line segment. According to the invention, a more reliable image edge cutting effect can be generated.

The invention belongs to the technical field of coal mine underground mining, and discloses a working face roof periodic weighting identification method and system based on a YOLO algorithm. The method comprises the following steps: mapping data into a three-dimensional space to generate a stope working face 3D stress nephogram, and generating a uniformly distributed and visualized stope three-dimensional grid stress distribution nephogram; the high-stress area, the low-stress area and the weighting area are marked with corresponding labels; performing real-time image recognition on the high and low stress areas by adopting a YOLO algorithm, and accurately positioning a roof weighting area; through an image edge detection technology, a time range of a roof weighting area and a corresponding hydraulic support distribution range are extracted, and intelligent analysis of area pressure change is realized. According to the method, image recognition algorithms such as YOLO and the like are combined, so that a coal mine does not depend on a traditional data processing mode any more, potential errors caused by manual intervention are reduced, data processing reliability and real-time performance are greatly improved, and safety monitoring efficiency and accuracy are improved.

The invention relates to the technical field of image processing, in particular to a method for judging the image blurring degree based on an airfield pavement FOD monitoring system. Comprising the following steps: selecting a reference image; edge detection and energy value calculation; collecting and marking image data; constructing a deep learning model; training a deep learning model; and evaluating and predicting. According to the design, before FOD detection is carried out, the fuzzy degree of the detected image is detected, and the image which does not affect foreign matter detection is screened out for subsequent operation; by understanding the application of the image blurring degree in FOD detection, the efficiency, accuracy and stability of a detection system can be remarkably improved, manual intervention is reduced, the automation level is improved, resource utilization is optimized, and a solid guarantee is provided for airport safety and operation.

The invention discloses a lightweight single-image super-resolution reconstruction method based on local and global feature collaborative enhancement perception, which is used for solving the problems of insufficient utilization of high and low frequency clues in single-image super-resolution, low feature fusion efficiency and difficulty in consideration of structural consistency and visual fidelity of reconstructed images. The network adopts a double-flow heterogeneous architecture; a local branch uses multi-type differential convolution explicit coding image edge and texture prior to enhance detail characterization capability; the global branch effectively models long-range dependency and low-frequency semantic information by integrating local, cross-regional and global multi-level spatial self-attention mechanisms. The frequency sensing fusion module provided by the invention generates a channel specific space weight based on frequency characteristics, self-adaptively fuses double-branch characteristics, accurately balances structure maintenance and detail enhancement requirements in a characteristic fusion process, and effectively reduces characteristic redundancy. According to the super-resolution network, the model complexity is remarkably reduced, high-resolution images which are consistent in structure, vivid in vision and rich in details can be generated while lightweight design is kept, and an effective scheme is provided for efficient and high-performance lightweight single-image super-resolution reconstruction model design.

The invention discloses a BEV perception optimization method based on semantic projection compensation and edge structure enhancement, and the method comprises the steps: constructing an image edge feature pyramid under convolution perception through an MSEII module, and enhancing the boundary significance of a weak texture region; and then image semantic embedding is extracted by using a deep semantic segmentation network, geometric projection is carried out based on camera-radar external participation internal reference, image semantic features are mapped to a point cloud space, and an SPE module is constructed to complete semantic completion and structure alignment. After the image and the point cloud features are fused in a unified BEV coordinate system, a spatial context relation is further modeled through sparse point convolution, and a three-dimensional target bounding box and a semantic tag are output through a decoupling detection head. According to the method, on the basis of guaranteeing model lightweight and deployment efficiency, the long-distance target detection capability and semantic consistency in a shielding scene are remarkably improved, and the method is suitable for multi-mode BEV perception application in other resource-constrained equipment such as a mobile robot.

The invention discloses a cloud type inversion method and system based on cavity window attention, and the method comprises the following steps: obtaining satellite remote sensing image data, extracting multi-dimensional spectral features, converting the latitude and longitude information of an image into spatial features through a position coding module, carrying out the feature splicing of the two features, and obtaining a fusion feature; a convolutional layer is input to extract basic features of image edges, textures and the like, a plurality of cavity window attention modules are stacked, each cavity window attention module comprises a cavity window division module and a weighted attention recovery module, the cavity window attention modules are used for extracting multi-scale image features, image resolution is recovered step by step through up-sampling, feature representation is refined, and a multi-scale image is obtained. According to the method, the features are mapped to the category space, the classification probability is output, cloud type inversion with higher accuracy and robustness is achieved, and an effective solution is provided for cloud information processing in the remote sensing image.

The invention discloses a micro-sensing trapezoidal method, device and equipment based on Aruco recognition and a medium, and relates to the field of computer vision. The method comprises the following steps: S1, projecting an Aruco grid and collecting an image, and extracting marked angular points to construct a position set; s2, calculating a trapezoid tension factor based on the trapezoid area and the opposite side distance, and quantifying local deformation; s3, fusing multi-labeled spatial parallax and tension information, constructing a continuous tension equilibrium domain, and generating a final mapping position through gradient correction basic inverse projection mapping and boundary disturbance compensation; and S4, outputting an inverse perspective matrix based on a fusion result, and estimating a microscopic pose offset parameter for fine adjustment of the projector. And through tension factor and spatial parallax modeling, adaptive correction of a distortion region is realized, the edge consistency and overall symmetry of the image are improved, and the stability and precision of the projected image are enhanced.

The invention discloses a camouflage target segmentation method based on multi-scale cross-spectral feature interaction, and belongs to the technical field of computer vision. The implementation method comprises the following steps: 1, constructing a camouflage target detection network; 2, obtaining visible light and infrared dual-mode features for the network subjected to feature extraction; 3, performing multi-layer feature fusion on the visible light image features and the infrared image features; therefore, bimodal complementary information extraction is realized; 4, constructing an edge sensing network for performing detail processing on the edge of the image so as to obtain an edge prediction result; 5, training the camouflage target detection network to obtain training parameters of the camouflage target detection network; optimizing training parameters of the camouflage target detection network; 6, inputting the image pair into the training optimization camouflage target detection network to obtain a predicted camouflage target detection result; compared with the prior art, the accuracy of boundary fuzzy recognition and appearance similarity recognition of camouflage target detection is improved.

The invention relates to the technical field of image edge analysis, in particular to an automatic leveling and tamping device for a road foundation. The method comprises the following steps: firstly, screening out all suspected uneven areas according to the edge distribution condition in a road foundation grayscale image; calculating brightness difference according to the characteristic that gray values of two side edges of a soil block generated when a road is uneven under the influence of illumination generally have large difference, so as to preliminarily screen out a preferable uneven area; then considering the characteristic that the edge of the soil block is generally not smooth enough, and determining the possibility of the interference region by combining the gradient amplitude distribution uniformity condition and the edge change stability condition of the edge pixel points of the optimized uneven region; therefore, a more accurate tamped uneven area is screened out according to the possibility of the interference area, so that the effect of leveling and tamping the road foundation according to the tamped uneven area is better.