73 results about "Edge enhancement" patented technology

The application belongs to the field of electric power system operation and inspection, and discloses an image super-resolution reconstruction method and related devices, based on an original image, a pre-trained image super-resolution reconstruction model is called to obtain a reconstructed image; the image super-resolution reconstruction model comprises an edge information pre-enhancement module, an attention module and a super-resolution reconstruction module; the edge information pre-enhancement module performs adaptive edge enhancement according to the uniformity of the original image to obtain an edge-enhanced image; the attention module comprises a serial edge attention module and a dynamic spatial attention module, which are respectively used for feature re-labeling of the edge-enhanced image in the channel dimension and the spatial dimension to obtain an edge-strengthened image and a globally optimized image; the super-resolution reconstruction module fuses the edge-enhanced image and the globally optimized image, and obtains the reconstructed image through channel dimension reorganization. The high-frequency details and edge structure of the image can be effectively restored, the clarity and fidelity of the reconstructed image are improved, and the quality of the reconstructed image is ensured.

The application discloses a kind of based on double filtering cooperation's remote sensing image scene classification method and system, in the method, for the problem that existing technology is difficult to give consideration to background noise suppression and key feature edge structure preservation when processing remote sensing image, a kind of double filtering cooperation optimization module is presented.The module in this paper introduces Gaussian filter smoothing channel to suppress unstructured high-frequency background noise, while introducing Gaussian Laplace edge enhancement channel to accurately capture and strengthen key geometric structure information such as feature contour;Through feature fusion, position coding, state space model and double activation gating mechanism, the multi-scale feature map is optimized and reconstructed to generate high signal-to-noise ratio and clear structure scene representation.The module is integrated into the remote sensing image scene classification model, and trained using focal loss, which can significantly improve the classification accuracy and robustness in complex texture interference scene.

This invention discloses a multi-scale, multi-semantic collaborative target detection method, system, medium, and product, belonging to the field of target detection. The method comprises: acquiring image data containing the target to be detected; performing multiple iterative feature processing on the object to be processed based on a preset detection model to obtain the final features; in each iteration, extracting initial features from the current object to be processed, grouping them into four sub-features along the channel dimension, and performing texture and edge enhancement, structural prior extraction, local context modeling, and global semantic representation generation through the four branches to obtain the features of each branch; fusing the features of each branch with the initial features to update the object to be processed; the object to be processed in the first iteration is the image data; after a preset number of iterations, performing target detection based on the final features and outputting the results. By implementing this invention, the problem of insufficient perception accuracy and robustness of existing target detection methods for multi-scale and multi-type targets in complex scenes can be solved.

The application relates to the technical field of image processing, and discloses a traditional Chinese medicine tongue image segmentation method for a mobile terminal, which comprises the following steps of performing feature extraction on a tongue image based on a tongue image segmentation network: obtaining first tongue body region features and second tongue body region features with different scales, performing edge enhancement on the second tongue body region features based on the first tongue body region features to obtain edge-enhanced second tongue body region features, performing global enhancement on the second tongue body region features through spatial displacement to obtain globally-enhanced second tongue body region features, and obtaining a tongue body region segmentation result based on the edge-enhanced second tongue body region features and the globally-enhanced second tongue body region features. Through the above scheme, accurate segmentation of a tongue body region is realized, the accuracy of tongue body boundary segmentation is improved, the segmentation stability is good, and the storage requirement and the calculation complexity of the tongue image segmentation network are low.

The application discloses a 2.5D MRI brain tumor segmentation method based on Mamba state space modeling and belongs to the technical field of medical image segmentation. The application aims at the problems of the existing brain tumor segmentation method, such as long-range dependence modeling deficiency, high calculation cost, limited boundary segmentation precision and poor adaptability of the Mamba architecture to the 2.5D segmentation scene, and a cross-slice priority traversal module suitable for the 2.5D slice space structure is designed, local details and global semantic features are balanced through a feature calibration and controlled fusion module, and a global perception and edge enhancement dual-decoder structure is constructed to decouple tumor global modeling and boundary modeling tasks. The application reduces the calculation complexity while ensuring the segmentation precision, solves the problems of tumor segmentation discontinuity and boundary blur and is suitable for clinical conventional image workstations and has a good application prospect.

The application discloses a kind of non-coherent full-color polarized edge imaging system and method based on reverse design photonic crystal, belong to optical imaging and computational optics technical field.System includes non-coherent light source module, polarization-dependent photonic crystal (PDPC) device, imaging detection module and image processing module;PDPC device is made of different all-dielectric material stack, with each layer thickness as trainable parameter, iteration optimization by gradient descent optimization algorithm, with ReLU function constraint lower limit of thickness, obtain optimal structure.Method by collecting orthogonal linear polarization state intermediate image, by spatial difference decomposition, using polarization difference optical transfer function satisfying first-order spatial differential condition, output edge enhancement image containing full color and polarization information.The application does not need coherent light source, it is stable in 400-700nm visible light band Work, high imaging efficiency, contrast ratio is optimal, provide feasible scheme for large-aperture broadband optical simulation processing.

This invention discloses a lightweight remote sensing small target detection method based on edge enhancement and multidimensional attention. First, multi-scale feature extraction based on stacked phantom-guided reparameterization is performed on the remote sensing image. Frequency-domain guided gated edge enhancement is employed, and global context modeling is performed on the enhanced feature map based on a multi-head self-attention mechanism. After residual connection and layer normalization, the enhanced feature map is fed into a feedforward network for nonlinear transformation, outputting globally encoded high-dimensional features. A target query based on self-attention information interaction is then performed, interacting the target query with the high-dimensional features based on cross-attention. Lightweight multidimensional attention is then used to perform channel and spatial weighting on the interacted high-dimensional features, mapping them to target category and bounding box coordinates via a prediction head, outputting the final detection result. This method addresses problems such as complex backgrounds, large target scale differences, low pixel ratio of small targets, and inherent speckle noise in synthetic aperture radar images.

This invention discloses a non-contact deformation monitoring system and method for rotary kilns. It includes: a pulsed laser emitting module for irradiating the operating rotary kiln cylinder with a time-domain encoded pulsed laser beam at a low swoop angle; an optical edge enhancement module for multiple reflections and refractions of the original contour light formed by the pulsed laser beam on the cylinder surface; a CCD synchronous acquisition module for receiving the contour light signal processed by the optical edge enhancement module and outputting a digital image; and an image processing and deformation calculation unit for performing edge extraction and reconstructing the current contour model from the digital image to obtain the cylinder deformation. An alarm is triggered when the deformation exceeds a set safety threshold. The entire system of this invention has no physical connection to the kiln body, completely avoiding the influence of high temperature and vibration on the sensors. Furthermore, it is designed to monitor small deformations in the early stages of material creep, supporting long-term tracking of slow deformations at the millimeter level, enabling pre-failure prediction, and improving safety and service life.

This invention provides a strawberry detection method based on an improved YOLOv5s framework, comprising: acquiring an image of a strawberry to be detected; inputting the image into a trained improved YOLOv5s detection network to obtain a detection result, wherein the improved YOLOv5s detection network includes an edge enhancement module in the backbone network and a focusing module in the neck network. The edge enhancement module effectively enhances the model's ability to perceive the features of the strawberry outline and the fine structure of the stem; the focusing module significantly improves the feature representation ability of small targets and occluded targets; enabling the model to more accurately identify strawberry fruits and stems in complex agricultural scenarios such as dense fruit and leaf occlusion, effectively reducing the false negative and false positive rates, and providing reliable technical support for automated agricultural harvesting.

This invention relates to the field of image enhancement technology, specifically to a method for edge sharpening and intracavitary enhancement of carotid artery plaque ultrasound images. The invention determines sampling points along the normal direction of the blood flow energy principal axis reference line; it reconstructs the features of each sampling point in the B-mode grayscale image and the Doppler energy image to construct a normal depth feature matrix; based on the feature differences between adjacent elements in this matrix, it performs probabilistic iteration and convergence to obtain a boundary steady-state probability distribution matrix; it uses the Doppler energy values ​​of the sampling points to correct the boundary steady-state probability distribution matrix and determine the junction points between the vessel wall and tissue; based on the positional distribution of the junction points between the blood flow energy principal axis reference line and the vessel wall and tissue, it performs region isolation and grayscale reconstruction on the B-mode grayscale image to obtain a B-mode reconstructed image. This invention extracts dual-modal features to construct a normal depth matrix, and through probabilistic iteration and Doppler energy extremum correction, it accurately locks the vessel wall junction by penetrating blood flow overflow artifacts, thereby improving the ultrasound image enhancement effect.

This invention relates to the fields of remote sensing image processing and agricultural information technology, and in particular to a method for refined extraction of winter wheat planting areas based on Gaofen-2 imagery and edge-enhanced DeepLabV3+. The method includes: acquiring Gaofen-2 imagery and performing preprocessing such as radiometric calibration, atmospheric correction, geometric fine correction, and image fusion to construct a labeled training dataset; constructing an edge-enhanced DeepLabV3+ network, which adds an edge enhancement module to the standard DeepLabV3+ network to explicitly learn the edge features of winter wheat fields and inject edge information into the decoder; designing a joint loss function, including segmentation loss, edge loss, and boundary-aware loss, and training the network end-to-end; and using the trained network to perform sliding window prediction and post-processing on the image to be extracted to obtain the refined extraction results of winter wheat planting areas. This invention significantly improves the segmentation accuracy of winter wheat field boundaries through the edge enhancement module and boundary-aware loss.

An adaptive multi-scale state-space model establishment method and its implementation in a perceptual scanning analysis method are presented. In multi-class pneumonia, especially in scenarios with class imbalance, small sample sizes, and blurred boundaries, it is difficult to simultaneously guarantee spatial structure modeling capability, pathological differentiation capability, and computational efficiency. There is a lack of standardized processing methods that address the detailed issues of spatial structure modeling capability, pathological differentiation capability, and computational efficiency. This invention divides the acquired and processed chest X-ray image into two data streams. One stream is processed through detection and adaptive multi-directional scanning to form a spatial model in the SSM state. The other stream is processed through multi-scale feature extraction to form an edge enhancement module and feature fusion data. The edge enhancement module, feature fusion data, and the SSM-state spatial model are interactively fused to form an interactive fusion model, which is then used to complete the classification output process.

This invention discloses a method for monitoring the particle size of ultrafine, high-fiber, fast-dissolving broccoli pollen based on image recognition, belonging to the field of immunoassay system technology. The method includes: constructing an image acquisition system; selecting an appropriate imaging module and shooting parameters based on whether the sample is broccoli pollen paste or dried powder; acquiring microscopic images of the sample; preprocessing the acquired microscopic images by sequentially performing image denoising, grayscale conversion, binarization, and edge enhancement to eliminate recognition errors caused by background interference and particle adhesion; using a contour extraction algorithm to identify particle contours in the preprocessed images, combining area, roundness, and pixel thresholds for feature filtering to remove impurities and pseudo-particle contours, obtaining a dataset of valid broccoli pollen particle contours; this invention achieves simultaneous analysis of particle size, particle size distribution range, and specific surface area, without requiring additional detection equipment, and can output multiple indicators directly related to broccoli pollen quality and process optimization from a single image acquisition.

This invention discloses a remote sensing change detection method based on edge-enhanced cross-attention and multidimensional loss. The method includes: inputting dual-temporal images into a trained Siamese backbone network to extract multi-scale backbone features from the dual-temporal images; inputting the multi-scale backbone features into a trained unsharpened mask cross-attention fusion model to extract spatial and semantic information from the dual-temporal images through cross-attention operations, and extracting edge information from the dual-temporal images through sharpening convolution kernels to obtain multi-level differential features at different scales; inputting the differential features into a trained multi-scale hierarchical connection decoder to obtain preliminary multi-scale prediction results for the changed region; and performing weighted fusion of the preliminary multi-scale prediction results to obtain the change prediction result. This invention can fully utilize the multi-scale information of the Siamese backbone network and improve the ability to extract edge information from dual-temporal images, thereby improving the accuracy of change detection.

This invention relates to the fields of medical image processing and auxiliary diagnosis of ophthalmic diseases, specifically to a method and system for retinal vessel and optic disc segmentation based on multi-dimensional attention and edge enhancement mechanisms. It first acquires and preprocesses fundus images, then constructs an EIAIU_Net segmentation model integrating a multi-dimensional attention module and an edge enhancement Transformer module. The model is trained based on a paired shuffling consistency strategy, which reduces the dependence on absolute positional information. After inputting the preprocessed image into the model, multi-path features are generated through residual fusion using the SSA-IAI module, and then bidirectional interaction between edge and global features is achieved using the EETF module's dual-branch structure to generate interactive enhancement features. Finally, pixel-level segmentation results are obtained through the output layer. This invention solves the problems of low efficiency and severe loss of detail in traditional methods, and insufficient segmentation accuracy and limited generalization ability of existing deep learning models, enabling automatic segmentation of retinal vessels and optic discs.

The application provides a long jump motion abnormality recognition method, system and device based on AI, relates to the field of artificial intelligence, and through collecting full-process video images of long jump motion, performing feature enhancement and edge enhancement on the images, detecting small part targets in the enhanced images based on an improved target detection algorithm, acquiring real-time coordinates and motion trajectories of the small part targets, and acquiring lower limb joint information in combination with human body posture estimation, core features of a take-off moment and a landing moment are extracted from continuous frame images, specific moments of the take-off moment and the landing moment are determined according to a judgment condition, abnormal behaviors are recognized according to the key moments, the real-time coordinates and the motion trajectories of the small part targets, and long jump rules are combined, and a recognition result is output. Small part target motion trajectories can be accurately captured, specific moments of take-off and landing can be determined, various long jump abnormal behaviors can be accurately recognized, and the problem of determination deviation caused by missed detection and misdetected small part targets can be solved.

This invention relates to the field of photolithography equipment control technology, specifically a method for pattern correction using unequal-distance triggering in photolithography machine scanning. The method includes: acquiring actual position data of the silicon wafer exposure field collected by a grating ruler at fixed intervals; inputting the data into a position error field model for analysis to generate a non-uniform position error cloud map; performing edge enhancement processing on the cloud map; extracting critical position points where the rate of change of position deviation exceeds a set slope threshold and dividing the error mutation interval; using this interval as a reference to set a variable delay to form an unequal-distance triggering control signal; adjusting the timing of exposure pulse release to compensate for motion deviations; simultaneously extracting high-frequency jitter components from the cloud map and correlating them with the position data of the error mutation interval to form a jitter feature packet; performing secondary correction on the trigger signal; and adding a small trigger advance at the interval boundary to offset the overshoot introduced by the position feedback delay. This method can improve the accuracy and adaptability of photolithography pattern correction.

This invention relates to the field of image processing technology, and more particularly to a method and system for assessing the quality of prefabricated construction based on image processing. The method includes the following steps: after removing the EVA transport protective padding that came with the prefabricated construction components, a panoramic image of the component splicing surface is captured using an industrial camera to obtain an original image containing the natural adhesion marks of the padding; multi-scale edge enhancement processing is performed on the original image to extract the contour data of the adhesion marks, and the global flatness of the component splicing surface is calculated based on the deformation features of the contour data; an industrial camera is used in a ring array to capture images of the pre-reserved grouting holes in the components, and the grouting hole boundaries are extracted after contour sharpening processing of the images; the three-dimensional axis of the grouting holes is constructed by combining the reflective features of the inner sidewalls of the holes. This invention achieves accurate detection and standardized assessment of the quality of prefabricated components through multi-polarized light sources and ring camera imaging technology, ultimately improving the accuracy of construction quality assessment.

A method for target detection in complex scenes captured by UAVs based on the D-FINE model is disclosed. The method includes: constructing a UAV aerial image target detection model by improving the D-FINE model; obtaining a trained UAV aerial image target detection model; obtaining the UAV aerial image to be detected; inputting the UAV aerial image to be detected into the trained UAV aerial image target detection model for target detection; and outputting the UAV aerial image target detection result. The advantages of this invention are: the LoG-Stem edge enhancement module combines traditional image processing operators with learnable deep features, effectively improving the ability to detect challenging targets; the complementary feature downsampling module generates more robust feature maps with complementary feature sets, enabling more accurate and robust analysis of remote sensing images; and the context awareness module enhances the model's ability to perceive key global contextual information. Overall, the method achieves improved real-time performance, increased detection accuracy, and enhanced comprehensive generalization ability with strong scene adaptability.

This invention proposes an image reconstruction method for physical degradation sensing of planar diffraction lens arrays. It utilizes a deep learning network integrating an optical degradation sensing mechanism to optimize sub-aperture feature fusion and weight allocation strategies, improving the imaging contrast and detail recovery accuracy of planar diffraction lens arrays. The method includes: acquiring a sequence of monochromatic sub-aperture images of the planar diffraction lens array carrying spatially non-stationary blur and diffraction degradation features; implicitly aligning the feature responses of each sub-aperture using an edge enhancement feature alignment module; constructing a pseudo-array feature representation; introducing a physically prior-guided weight modulation strategy during adaptive grouping upsampling, learning the distribution law of spatially heterogeneous point spread functions through degradation sensing units, and spatially adaptively modulating the feature contribution; and finally reconstructing a high-quality grayscale image. This invention significantly suppresses the diffraction blur and low contrast phenomena unique to planar diffraction lenses, better sensing optical degradation distribution and preserving high-frequency details in the image. It also provides a new approach for subsequent joint reconstruction research of arrayed lightweight imaging systems.

This invention discloses a fast autofocusing method for layered imaging based on vortex edge enhancement and selective linear fitting, comprising: acquiring the initial complex amplitude of an object; propagating the initial complex amplitude to a virtual Fresnel zone plate, modulating it to generate a hologram, extracting edge information during the backpropagation of the hologram, locating the region of interest (ROI), and calculating the sharpness bias within this ROI, storing it along with the corresponding axial distance in a historical dataset; during iterative iteration, substituting the updated distance parameters for each iteration of the layered imaging algorithm, classifying all data points in the historical dataset into linear or saturated intervals, performing least-squares fitting on data points classified as linear intervals, outputting the intercept and slope of the fitting function, and calculating new axial distances based on the current data point classification using different strategies, before proceeding to the next iteration. This invention enables high-precision and high-efficiency automatic calibration in a single preprocessing step.

The present application relates to the technical field of image processing, in particular to a fisheye lens image pixel-level edge enhancement and denoising method, comprising: a data acquisition step: acquiring original fisheye image data and preset fisheye lens optical distortion parameters; a weight generation step: determining the distortion stretching rate of a pixel point according to the optical distortion parameters; and generating a spatial density weight map; a threshold determination and denoising step: determining a target denoising threshold according to the spatial density weight map; denoising the original fisheye image data to generate an intermediate denoising image; a convolution kernel generation step: generating an adaptive curved surface convolution kernel according to the optical distortion parameters; an edge enhancement step: using the adaptive curved surface convolution kernel to perform edge enhancement processing on the intermediate denoising image; and generating an enhanced image; the present application effectively avoids the risk of irreversible erasure of key semantic features in the edge field, and ensures high-reliability feature preservation in the extreme field.

The application discloses a two-dimensional omnidirectional edge detection phase correction automatic optimization method based on a double-layer titanium dioxide metasurface, and belongs to the technical field of metasurface light field regulation and optical image processing. In view of the problems of unstable edge enhancement effect, insufficient adaptability and evaluation susceptible to local abnormalities of an existing scheme, a candidate model set containing an original scheme and various radial phase correction schemes is established on a basic radial phase distribution, the original scheme is used to determine main edge reference positions in multiple fixed directions, each candidate model extracts an edge response near the position, an average signal-to-noise ratio and a full width at half maximum are calculated in multiple directions, and then a unified comprehensive score is calculated, and the candidate model with the highest score and parameters thereof are output as an optimal scheme. The method can be used for high-speed low-power optical analog edge detection and integrated optical image processing, and improves the adaptability and robustness of edge detection.

This invention relates to a deep learning-based method and medium for recognizing VIN codes on vehicle body stamps. The constructed character recognition model includes: an input layer for receiving VIN code images; a feature extraction layer for performing multi-branch convolutional fusion and downsampling on the VIN code image to obtain a main feature map, and then performing edge enhancement and sequence modeling on the main feature map to obtain an enhanced feature map and character space features; a feature fusion layer for performing residual fusion on the main feature map, enhanced feature map, and character space features to obtain fused features; and an output layer for outputting VIN code characters based on the fused features. This provides a novel recognition technology solution that can adapt to complex working conditions, fully utilize the structural characteristics of VIN codes and the logical relationships between characters, and possesses high accuracy and high robustness.

The disclosure provides an atomization state recognition method and device of an atomizing nozzle and a training method of an atomization state recognition model, which can be applied to the technical fields of image processing and computer vision. The atomization state recognition method of the atomizing nozzle comprises: performing droplet edge enhancement processing on an original atomization image generated by the atomizing nozzle in a production process to obtain a target atomization image without imaging noise; performing multi-dimensional asymmetric convolution processing on the target atomization image to obtain multi-dimensional atomization features; performing weight mask processing on the multi-dimensional atomization features based on a spatial attention mechanism to obtain spray profile features with droplet boundary information and spray cone structure information, and performing weight fusion processing on the spray profile features based on a double-head attention mechanism to obtain atomization fusion features; and performing classification processing on the atomization fusion features to obtain an atomization state recognition result of the atomizing nozzle.

The present application proposes a wafer epitaxial film thickness measurement method and system based on trajectory adaptation, belonging to the field of measurement and testing and semiconductor manufacturing technology. The measurement method includes wafer feeding and preliminary transfer steps, wafer positioning and correction compensation steps, measurement path planning steps, film thickness data acquisition and processing steps, data comparison and trend report generation steps, and abnormal alarm triggering steps. The measurement system includes feeding and transferring units, positioning and compensation units, measurement path planning units, film thickness data acquisition and processing units, data comparison and trend report generation units, and abnormal alarm triggering units. The positioning and compensation unit includes a correction mechanism and a compensation mechanism. The present application improves the precision and efficiency of wafer epitaxial film thickness measurement through trajectory adaptive planning and edge enhancement algorithm; combined with cloud collaborative analysis, it realizes real-time abnormal alarm and trend prediction, which helps to improve the yield and intelligent level of semiconductor manufacturing.

This invention discloses an intelligent detection method for surface defects on silicon carbide wafers. A detection network is constructed based on an improved YOLOv12 architecture. A dual-branch backbone network is used to extract the geometric morphology and lattice texture features of the defects, which are then fused using an adaptive feature fusion module. The neck network employs a four-scale progressive feature pyramid structure, adds a P2 ultra-high resolution detection layer, and embeds a hybrid attention enhancement module. During training, a multi-task joint loss function with edge enhancement loss and a multi-stage progressive training strategy are used, combined with adaptive dynamic anchor boxes and multi-scale test fusion inference to achieve high-precision detection. This invention addresses the defect characteristics of SiC epitaxial layers, improving the detection capability for micro-defects and transparent defects, while balancing real-time performance and generalization.

The application discloses a railway track instance segmentation model establishing method and device, equipment and a storage medium, relates to the technical field of computer vision and intelligent image processing, and comprises the following steps: acquiring railway track image data, and pre-processing and data enhancement are performed on the image to obtain enhanced track training samples; an edge enhancement algorithm based on a Scharr operator is introduced in the data enhancement stage to strengthen the track region boundary features; an improved YOLO11n-seg instance segmentation network structure is constructed, the network comprises a double-path multi-branch feature pyramid DMBFPN module, a multi-scale edge feature fusion module MSEFF and a decoupled efficient edge perception detection segmentation head DEED-Seg; the enhanced samples are input into the network for training to obtain a railway track instance segmentation model; and the to-be-detected image is input into the model to output an instance segmentation result. Through the fusion of the edge enhancement and the multi-scale feature fusion strategy, high-precision railway track instance segmentation is realized under low calculation amount, and the method has the advantages of high precision, strong real-time performance and easy deployment.

This invention discloses a method, apparatus, server, and medium for detecting small targets based on feature reconstruction, belonging to the field of target detection technology. It includes: inputting primary features obtained from image convolution into a two-layer DRSS-Unit module for processing to obtain shallow edge enhancement features; inputting the shallow edge enhancement features into an MGSP-Unit module for processing to obtain mid-level semantic transition features; inputting the mid-level semantic transition features into an MGSP-Unit module for processing to obtain deep sparse semantic features; inputting the deep sparse semantic features into an SDF-AIFI module for processing to obtain frequency domain global enhancement features; inputting the frequency domain global enhancement features, shallow edge enhancement features, and mid-level semantic transition features into a CCFM module for feature reconstruction to obtain reconstructed fusion features; and inputting the reconstructed fusion features into a decoder to obtain the small target detection result. By constructing a cascaded feature reconstruction and enhancement module, progressive reconstruction and enhancement of multi-level features are performed to achieve the detection of small targets in complex scenes.

This invention discloses a computer vision-based method and system for automatic empty can counting and warehousing verification. By specifically addressing the image reflection interference problem in the complex environment of delivery stations, this invention performs reflective region segmentation and compensation processing on the images of empty cans, effectively eliminating the negative impact of reflections on the smooth surface of the cans on contour recognition. Furthermore, adaptive edge enhancement is applied to the de-reflected images, significantly improving the extraction accuracy of edge features for each empty can. Then, while accurately identifying the edge images of each empty can, brand label classification is completed, allowing for the determination of the quantity of empty cans from different brands at once, which can then be directly compared and verified with SaaS deposit invoices. Thus, this invention completely replaces the inefficient traditional method of manual counting, identification, and data entry, significantly improving the efficiency and accuracy of empty can warehousing and achieving automation and intelligence in the empty can warehousing process.