49results about How to "Accurate focus" patented technology

The invention belongs to the technical field of software engineering, particularly relates to an automatic driving system performance enhancement method based on an expert hybrid architecture, and aims to solve the core problems that an end-to-end automatic driving system is confronted with semantic fuzziness to cause unreliable decision, multi-task interference hinders optimization planning, too long reasoning delay increases driving risks and the like. According to the method, an ExpertAD framework is provided, task key features are amplified through a perception adapter (PA), and the relevance of scene context understanding is guaranteed; related driving tasks are dynamically activated through a sparse expert mixture (MoSE), and task interference is minimized; and in combination with a customized training loss function, collaborative optimization of planning effectiveness and reasoning efficiency is realized. Experiments show that compared with an existing method, the method has the advantages that the average collision rate is reduced by 20%, the reasoning delay is reduced by 25%, higher multi-skill planning capacity is achieved in rare scenes (such as accident handling and first-aid vehicle avoiding), and good generalization is achieved for unseen urban environments.

The invention discloses an automatic focusing method and device, computer equipment and a computer readable storage medium, and relates to the technical field of focusing. The method comprises the following steps: performing multi-target tracking detection on a preview image frame to obtain a face frame set, a human eye frame set and a multi-target tracking parameter; performing scale measurement aperture unification on the face frame set and the eye frame set to obtain a target face frame width and a target eye frame width; calculating a human face available threshold value and a human eye available threshold value; determining a depth-of-field coverage judgment result of the focusing target according to a depth map corresponding to the preview image frame; and according to the target face frame width, the target eye frame width, the multi-target tracking parameter after smoothing normalization, the depth-of-field coverage judgment result, the face available threshold and the eye available threshold, determining a focusing strategy to perform focusing switching. Therefore, the stability of the exported picture, the accurate focus and the smooth switching are ensured through automatic focusing.

The invention relates to an acoustic signal processing and biological recognition technology, in particular to a method for fusing sound source localization and voiceprint recognition, a chip and electronic equipment. The method comprises the following specific steps: synchronously acquiring an environment audio signal through an audio acquisition module array and processing the environment audio signal to obtain audio data; and analyzing the audio data to obtain deep fusion voiceprint features. And judging whether a registered user feature library contains the deep fusion voiceprint feature or not, if so, outputting a sounder identity, and if not, marking and endowing an identification number of the sounder identity. And positioning the sound source based on the audio data to obtain a space coordinate thereof. And establishing an identity-position association model, binding identity information output by voiceprint recognition with space coordinates obtained by sound source positioning, and generating identity-position association data. According to the method, the technology is fused and innovated, accurate linkage of identity-position is realized, high real-time performance and strong anti-interference capability are realized, and the method is adaptive to complex scenes.

The application discloses a deep-sea biological image processing method based on a neural network, belongs to the technical field of image processing, and is used for deep-sea biological image processing and comprises the following steps: preparing a target data set, constructing a neural network, inputting a processing result of a Backbone layer into an improved Neck layer, obtaining a deep-sea biological image processed and optimized after being processed by the improved Neck layer, and inputting the deep-sea biological image processed and optimized into a Detect layer; the improved Neck layer comprises a dynamic self-adaptive up-sampling module, a multi-level feature fusion module, an HT_C2f module based on HyperTransformer, and a down-sampling convolution module. The application realizes accurate focusing of target features under a low signal-to-noise ratio condition by improving the Neck layer, improves the feature consistency of multi-scale target processing, reduces the loss of key information in the dimension reduction process while improving the small target feature retention rate.

The utility model provides an optical fiber head adjusting device and an optical fiber head adjusting system. The optical fiber head adjusting device comprises a horizontal adjusting piece and a vertical adjusting piece. The horizontal adjusting piece comprises an optical fiber fixing section and a horizontal adjusting section and is provided with a first through hole. The vertical adjusting part comprises a containing section, a vertical adjusting section and a positioning section, threads are arranged on the outer side of the vertical adjusting section, and the positioning section is cylindrical. The vertical adjusting part is provided with a second through hole and a third through hole, the center lines of the second through hole and the third through hole coincide, the diameter of the second through hole is larger than that of the third through hole, and therefore a step structure is formed at the junction of the containing section and the vertical adjusting section, and the horizontal adjusting part is placed on the step structure. A plurality of horizontal adjusting holes are formed in the side wall of the containing section, and horizontal adjusters are arranged in the horizontal adjusting holes correspondingly and used for pushing the horizontal adjusting pieces in different directions. According to the technical scheme, the position of the optical fiber head can be adjusted in the horizontal plane and the vertical direction, and therefore it is effectively ensured that the transmission path of optical signals is accurate.

This invention provides an intelligent analysis system for photoelectric detection spectral data based on deep learning, belonging to the fields of photoelectric detection, spectral data analysis, and deep learning application technology. The system includes core units such as a spectral data acquisition unit, a preprocessing unit, a feature extraction unit, and a deep learning analysis unit, as well as auxiliary units for anomaly detection and data storage, forming a closed-loop architecture of acquisition-processing-analysis-output-optimization. It improves data quality through step-by-step preprocessing algorithms, reduces computational load by adaptively selecting key features, deeply mines spectral patterns using a model that integrates convolutional neural networks and attention mechanisms, achieves rapid scenario adaptation through transfer learning, and dynamically optimizes model parameters based on a feedback mechanism. The system solves the problems of low accuracy, poor adaptability, and long training cycles in traditional spectral analysis, achieving accurate, efficient, and intelligent spectral data analysis, and is suitable for various photoelectric detection scenarios such as qualitative and quantitative analysis of substances and environmental monitoring.

The application discloses a vibration wave measuring system for traditional Chinese medicine shallow needle manipulation analysis, and belongs to the technical field of vibration wave measurement, and specifically comprises the following steps: a first beam splitter in a laser Doppler vibrometer is used to divide outgoing laser into a measuring light beam and a reference light beam, the measuring light beam is vertically focused on the surface of a needle handle, the reference light beam is vertically focused on an adjacent unoperated skin area, after the two light beams are reflected, interference occurs when the light beams return to a probe to form a first interference light signal, meanwhile, a part of the returned reference light beam is separated to interfere with local light of the system to form a second interference light signal, after photoelectric conversion, common-mode background noise is eliminated through a differential amplifier, and finally, a pure needle handle vibration wave is output, and high signal-to-noise ratio non-contact measurement of high-frequency weak vibration signals generated by traditional Chinese medicine shallow needle manipulation is realized.

The utility model relates to an automatic focusing three-dimensional five-axis laser cutting head which comprises an optical fiber QBH automatic collimation assembly, a C-axis rotating assembly, an A-axis rotating assembly, a light path reflection assembly and a cutting assembly, and has the advantages that the structure is compact, focusing is accurate, the optical fiber QBH automatic collimation assembly and an automatic focusing lens assembly are adopted, the center position and the focal length of a light beam can be accurately adjusted, and the cutting efficiency is improved. Stable and consistent cutting quality is guaranteed, and cutting precision is improved. The protection assembly is internally provided with pressurized gas to prevent dust pollution, and the service life of the optical element is prolonged; the anti-collision assembly is in magnetic attraction connection with an annular magnet through a contact type sensor, the cutting head is effectively protected against collision damage, the equipment reliability is improved, and the service life is prolonged. The laser machining device is high in adaptability, can adapt to various types of lasers, meets different machining requirements, has good universality and flexibility, and expands the application range. The automatic focusing function reduces manual intervention, accelerates the cutting speed, improves the production efficiency and reduces the production cost.

This invention provides a high-heat-dissipation battery component for new energy vehicles, belonging to the field of new energy vehicle heat dissipation battery technology. It includes a protective shell, inside which a cooling module is installed. The cooling module is used to cool the entire battery module. The battery module consists of a mounting box and battery cells. The battery cells are installed in mounting slots within the mounting box. A top plate is installed on the top of the battery module. The cooling module is connected to a serpentine cooling pipe, which has several adjusting components. Several adjusting components are arranged between adjacent sections of the serpentine cooling pipe. This invention, through the corresponding combination design of the cooling module, the serpentine cooling pipe and adjusting components, and the serpentine cooling pipe and adjusting components, achieves dual heat dissipation functions: overall cooling of the battery module and precise heat dissipation of individual battery cells, thus improving maintenance convenience.

The present application relates to a pathological image cell segmentation method based on a U-Net model, belonging to the field of deep learning and medical image segmentation. The present application adopts an improved U-Net model, and in the preprocessing stage, converts the original point label data into a continuous density map. A multi-scale attention module is added in the decoder stage of the network. The module effectively combines multi-scale feature fusion and soft attention mechanism, can comprehensively capture global and local features in the image, and significantly enhance the representation ability of key cell features, thereby improving the accuracy of cell segmentation. In addition, the present application adds a pixel channel fusion module between each layer of the decoder, which further enhances the model's ability to identify details by integrating local information and pixel-level attention into the channel attention mechanism. The present application can effectively segment Ki-67 positive cells, Ki-67 negative cells and tumor infiltrating lymphocytes in IHC stained pathological images.

The utility model provides a scraper fineness meter calibrating device which comprises a calibrating workbench, a calibrating jig, a scraper fineness meter, a driving case, a calibrating screw rod, a driving motor and an identifying structure, the calibrating jig is installed at the front end of the surface of the calibrating workbench, the scraper fineness meter is placed on the calibrating jig, and the driving case is installed on the calibrating screw rod. The driving case is mounted at the rear end of the surface of the verification workbench; the verification screw rod is mounted in front of the driving case, the driving motor is fixedly mounted at the top of the driving case, and an output shaft of the driving motor is connected with the verification screw rod; the identification structure is slidably mounted in front of the driving case and is engaged with the verification screw rod; according to the utility model, with the arrangement of the verification tool and the identification structure, a large amount of workers are not required to participate in verification, the accuracy of obtained information is high, and the verification process is simple.

The application provides a mask base cloth and a preparation method and application thereof, and the preparation method comprises the following steps: a first mixed fiber comprises a combination of fine denier viscose fibers and modal fibers, which is subjected to opening, double-carding and laying, to obtain a skin-felt fiber web layer; a second mixed fiber comprises a combination of vinylon fibers and seaweed fibers, which is subjected to opening, double-carding and laying, to obtain a micro-elastic web layer; the skin-felt fiber web layer and the micro-elastic web layer are stacked and then subjected to water jet entanglement, and the obtained semi-finished product is dried to obtain the mask base cloth. Through the design of fiber types and the compounding of the fiber types with a double carding machine, a specific double-layer structure is obtained, and the water jet process is combined, so that the mask base cloth has a specific microporous structure, cavitation effect and energy loss are avoided, the mask base cloth has excellent skin feeling, high air permeability, excellent liquid absorption and retention capacity, mechanical properties and acoustic properties, can be highly matched with ultrasonic beauty equipment, and can play a lifting and tightening effect.

This application is applicable to the technical field of intelligent terminals, and provides an electronic rearview mirror image stuttering processing method, device, terminal device and storage medium. The method includes: during the video stream acquisition process, marking the image data of a preset pixel region of each frame of image to generate a marked video frame containing marked blocks, where the marked blocks change with the frame; during the video stream display process, extracting the display data of the marked blocks from the marked video frame; based on the display data corresponding to the marked blocks of multiple consecutive marked video frames, determining whether the video stream has stuttered; if stuttering occurs, performing a fault recovery operation. This application can achieve precise detection and rapid recovery of image stuttering, improve the reliability of the electronic rearview mirror system, and thus provide a strong guarantee for driving safety.

In order to solve the problems of low evaluation efficiency and poor accuracy in the prior art, the invention provides a camera module multi-focal-length evaluation method and system, and the method comprises the steps: installing a camera module and aligning a test device, controlling the focal length of the camera module to be switched to different focal lengths, starting a self-adaptive light source adjustment system, and adjusting the focal lengths of the camera module to different focal lengths. A light source is automatically adjusted according to the change of ambient light, and a computer system captures images of all focal segments, analyzes image data to calculate analytic power and outputs a test report. According to the invention, through cooperative work of the components, rapid and accurate focusing and automatic evaluation under a multi-focal-section condition can be realized.

The utility model belongs to image equipment technical field discloses an independent anti -shake module, carrier and support movable connection, base and support are connected through anti -shake elastic sheet, one of carrier and support is provided with focusing coil, the other is provided with focusing magnetite, anti -shake elastic sheet has first connecting part, second connecting part, third connecting part, second connecting part is located between first connecting part and third connecting part, first connecting part and second connecting part are connected with base, third connecting part is connected with support, second connecting part and base have gap one between, point damping glue at gap one to make second connecting part and base connect. The utility model discloses with anti -shake elastic sheet can effectively avoid the deformation or fracture of the suspension wire after long -term use and cause the decline of anti -shake ability, is favorable in guaranteeing the anti -shake performance while, improves the life. The utility model discloses still a kind of electronic equipment with the independent anti -shake module of above.

The application discloses an automatic focusing method, system and medium, the automatic focusing method comprises the following steps: identifying the magnification of an objective lens, screening the defocus threshold corresponding to the magnification; constructing the theoretical motion model, the measured motion model and the measured variance of the focusing process parameters in the focusing process, and predicting the focusing process parameters at the next moment; analyzing the deviation between the measured value and the predicted value of the focusing process parameters at the same moment, determining whether the deviation of at least one focusing process parameter is greater than the set deviation threshold, if there is, dynamically adjusting the theoretical motion model; if not, driving the objective lens to focus when the predicted defocus is less than the defocus threshold. The application determines whether to dynamically adjust the theoretical motion model by the deviation of the predicted and measured focusing process parameters, eliminates the error, realizes fast focusing, meets the focusing requirements of objective lenses with different magnifications, and improves the focusing efficiency.

This invention relates to the field of image processing technology and discloses a focusing method for an endoscope system and an endoscope system. The method includes: step S1, acquiring the current image captured by the endoscope at the current focal length; step S2, determining the candidate focusing region on the current image; step S3, determining the sharpness evaluation value of the candidate focusing region; step S4, determining whether the sharpness evaluation value meets the preset sharpness condition; if not, changing the focal length of the endoscope and returning to step S1; if yes, determining that the endoscope system has completed focusing. This invention effectively addresses the problem of incorrect focusing, not only achieving more precise focusing and ensuring the acquisition of high-quality images, but also improving focusing efficiency, reducing the risk of blurring, and effectively enhancing the safety and effectiveness of medical operations.

This invention discloses a multi-view few-sample Android malware classification method based on optimal guidance matching, belonging to the field of information security technology. The invention includes constructing multi-view grayscale images, training a backbone network, optimal guidance matching classification, and dynamic fusion of multiple views. First, the method extracts permissions, APIs, components, and intent features of the Android application to construct multi-view grayscale images. Then, it trains a backbone network that integrates attention mechanisms and self-supervised rotation prediction to extract discriminative features with geometric structure awareness. Valid guidance samples are identified through optimal guidance matching, and category similarity scores are calculated. Finally, adaptive weights are generated based on view confidence, and the multi-view scores are dynamically fused to complete the classification. This invention improves the accuracy and robustness of malware family classification in few-sample scenarios, reduces noise interference, and is suitable for rapid and accurate identification of malware.

The present application relates to the technical field of steel pipe defect detection, and particularly relates to a defect identification method for steel pipe manufacturing, comprising: acquiring steel pipe surface weld image data and determining a weld appearance index according to the image data, wherein the image data comprises a weld width, a weld excess height and a undercut depth; determining the weldability of the steel pipe according to the weld appearance index, determining the reason for unqualified welding according to the average undercut depth of the weld under the condition of unqualified welding; determining an ultrasonic detection strategy according to the defect corresponding rate of the weld width and the weld excess height under the condition of qualified welding, and acquiring a defect characteristic value inside the weld through ultrasonic detection; determining the weldability of the steel pipe manufacturing according to the defect characteristic value, and determining whether the steel pipe meets the preset standard according to the defect area proportion under the condition of unqualified steel pipe manufacturing. The present application improves the accuracy and efficiency of steel pipe defect identification.

The application relates to the technical field of data optimization, and discloses a SQL optimization method based on dynamic statistical feature perception, equipment and a medium. The method comprises the following steps: identifying and analyzing a to-be-optimized SQL, and extracting table fields in filtering, association, grouping or sorting conditions as predicate columns; calculating the selectivity of each predicate column, and determining a predicate column with selectivity lower than a threshold value or a single value proportion exceeding a preset proportion as a key column; extracting column-level statistical information including histogram boundary values, null value proportions, different value quantities and the frequency of the most common values for the key column, and converting the column-level statistical information into a natural language statistical summary through a statistical feature descriptor; constructing a prompt word context based on an original SQL statement, table structure information of the key column and the summary, and inputting a large language model to generate a candidate SQL. According to the application, key columns are dynamically screened, and a natural language description of data distribution perception is generated, so that the consumption of Token input of the large language model is reduced, irrelevant information interference is avoided, and the SQL optimization precision is improved.

This invention belongs to the field of target detection technology, specifically relating to a multimodal feature fusion method for water target detection. The method includes the following steps in sequence: a preprocessing step, including: based on two adjacent RGB image frames as input, separating mirror perturbations to generate a mirror mask and a motion cue map; an alignment and fusion step, including: using a stable frame as the primary modality and the mirror mask and motion cue map as secondary modalities, extracting primary and secondary modal features; and inputting the fused features into a detection head to obtain initial candidate boxes and category scores; and a loop closure determination step, including: based on the conformal saliency ratio calculated for the initial candidate boxes and calibrated within the same frame, outputting the final detection result for that frame. This invention can automatically identify and eliminate unstable candidates caused by reflection. It also enhances the robustness and consistency of the model in complex scenes such as strong reflections, water wave disturbances, and changes in viewing angle.

The invention discloses a continuous stirring reaction kettle operation state monitoring method based on time sequence data analysis, and relates to the technical field of process state monitoring, and the method comprises the following steps: collecting multi-source operation time sequence data comprising stirring motor driving signals, material components, temperature and pressure, and product quality; performing cross-domain correlation analysis on the driving signal and the material components, and identifying a coupling correlation mode; dynamic process segmentation is performed on the temperature and pressure data based on the mode, and feature fragments of all reaction stages are extracted; constructing a causal inference network model between the feature fragments and the product quality; reversely tracing key process intervals and features influencing quality through the model, and generating a process traceability report; and judging the comprehensive operation state of the reaction kettle. According to the method, the accuracy and interpretability of state monitoring are improved by revealing process internal association and realizing causal traceability.

The application relates to the technical field of computer vision and deep learning, in particular to a multi-scale edge information multi-target detection method fusing text, which comprises the following steps: constructing a basic YOLO-World model comprising a backbone network, a neck network and a head network, and optimizing the structure of the basic YOLO-World model, including: replacing an original C2f module with an MSES module in the backbone network; introducing an AFCAtion attention module at the end of the backbone network; improving the feature fusion mode of an original feature pyramid network by adopting an SSFF module in the neck network; training the optimized model to obtain a target detection model; using the target detection model to perform target detection on an input image and outputting a target detection result; and solving the problems that an existing text fusion method is difficult to efficiently process the semantic difference between an image and text when facing a complex background, and the combination of an image and text is still insufficient for understanding of part features.

This invention discloses a video prediction model compression method and system based on optical flow-guided knowledge distillation, belonging to the field of computer vision technology. The method includes: generating optical flow distillation loss by constraining the optical flow prediction distribution of the student network to be consistent with that of the teacher network; generating channel alignment loss by aligning the channel attention distributions of the teacher and student networks using divergence; generating pixel-level reconstruction loss based on the difference in pixel values ​​between the generated image predicted by the student network and the real image; generating perceptual loss based on the difference in high-level semantic feature space between the generated image predicted by the student network and the real image; and obtaining the trained student network based on the optical flow distillation loss, channel alignment loss, pixel-level reconstruction loss, and perceptual loss. This invention can solve the performance degradation problem caused by neglecting spatiotemporal characteristics in existing compression methods for video prediction tasks.

The invention belongs to the technical field of artificial intelligence, and relates to an artificial intelligence-based image analysis method, which comprises the steps of performing feature extraction on a first image and a second image, and obtaining a first feature pyramid and a second feature pyramid; performing semantic segmentation to obtain a first semantic probability graph and a second semantic probability graph; performing attention generation on the first semantic probability graph and the second semantic probability graph to obtain a first attention weight graph and a second attention weight graph; respectively generating a first foreground exclusion confidence coefficient and a second foreground exclusion confidence coefficient based on the first semantic probability graph and the second semantic probability graph; and performing similarity calculation on the first feature pyramid, the second feature pyramid, the first attention weight map, the second attention weight map, the first foreground exclusion confidence coefficient and the second foreground exclusion confidence coefficient to generate background similarity. The method can be applied to image similarity analysis scenes in the financial science and technology field and the digital medical field, and the accuracy of image background similarity calculation is improved.

The application discloses a telescopic lens connecting structure, a lens driving device, a photographing device and an electronic device, and technical scheme points are as follows: a telescopic lens connecting structure comprises a base, a telescopic shell and a telescopic component; the telescopic shell drives the telescopic component to move along the direction of the lens optical axis; the telescopic shell is provided with a surface flat plate for limiting the telescopic component from separating from the telescopic shell, and the surface flat plate is located at one end of the telescopic shell away from the base; a guide assembly for guiding the telescopic component is arranged between the telescopic shell and the telescopic component; and the telescopic component is provided with a protection structure. The protection structure plays a protection and reinforcement role, improves the connecting strength, and has the advantages of simple structure, easy assembly and cost saving; and the problem that the light quantity of an existing mobile phone over 5 times long-focus periscopic camera is weak and better image quality cannot be obtained is solved.

The application belongs to the technical field of image processing, and particularly relates to a numerical control machine tool cutter wear recognition method and system based on image processing, comprising the following steps: acquiring a cutting edge image of a cutter to be recognized, and inputting the cutting edge image into a pre-trained boundary perception refinement network; the boundary perception refinement network extracts a multi-scale feature map through an encoder, and utilizes an initial prediction module to generate a global probabilistic initial wear area mask and an initial boundary direction field in parallel according to the highest layer semantic features output by the encoder; and a semantic guided multi-scale directional boundary feature map and a boundary clear wear area mask are generated according to the probabilistic initial wear area mask and the initial boundary direction field. The application has good stability to illumination changes, oil stain interference and geometric deformation of wear morphology in the cutting edge image, and improves the reliability of cutter wear recognition.

This utility model discloses a device for visual inspection of lenses. The key technical features include a machine platform on which multiple industrial cameras for visually inspecting lens appearance and a light diffuser are placed. A feeding plate is also provided on the machine platform, holding several trays of lenses to be inspected and trays of defective lenses. The machine platform also includes a driving device for moving the feeding plate along the X and Y axes to inspect the lenses, and a suction pen for picking up defective lenses and placing them into the defective lens tray. This utility model enables automated visual inspection by moving the lenses instead of the industrial cameras, avoiding the impact of camera movement on inspection accuracy and greatly improving inspection efficiency and quality.

This invention discloses a microscopic imaging focusing method and an automatic focusing device, belonging to the field of microscopic optical imaging technology. The microscopic imaging focusing method sequentially includes: preparation, coarse focusing, and fine focusing; the preparation includes calibrating the relationship between the image sharpness of chip markers and the driving current of the liquid lens. This invention is applied to the imaging focusing of complex biological samples containing a large number of impurities, solving the technical problem that existing automatic focusing methods cannot achieve fully automatic, high-speed focusing on complex biological samples within microfluidic chips, enabling fully automatic and rapid focusing for highly complex biological samples.

The invention relates to the technical field of noninvasive medical imaging, acousto-optic coupling and targeted drug delivery, and discloses a lesion finding and positioning and targeted drug guiding system and method based on an acousto-optic coupling functional film layer. Based on intelligent processing equipment with acoustics, optics and signal processing capacity, non-invasive discovery and accurate fault positioning of human tissue / body fluid lesions and directional guidance and local enrichment of targeted drugs are achieved in a time-sharing mode through the same physical area of the same acousto-optic coupling integrated functional film layer. The membrane layer multiplies the frequency of a low-frequency signal of 20Hz-20kHz into medical ultrasound of 2MHz-10MHz to complete imaging positioning, and meanwhile, targeted drug guidance such as optical / acoustic power, magnetic targeting and ultrasonic transdermal infiltration promotion is realized through the special optical / energy conducting layer, so that the drug focus enrichment rate is increased, and the side effects of the whole body are reduced. Equipment hardware does not need to be transformed, the film layer is flexible in form and wide in adaptability, medical diagnosis and treatment functions are integrated into consumer electronic equipment for the first time, low-cost popularization is achieved, and the system is suitable for targeted drug treatment guidance of various lesions such as solid tumors, virus infection lesions and chronic inflammation lesions, can be applied to multiple scenes such as families, primary medical treatment and hospitals and has wide application prospects. And the method has the remarkable advantages of high precision, noninvasive safety and the like.