303 results about "Image structure" patented technology

The invention discloses an automatic software function testing and evaluating method based on a multi-modal large model, and particularly relates to the field of non-embedded software function testing and evaluating, which comprises the following steps: collecting software texts, images and structured data, generating corresponding feature vectors, aligning through a cross-modal comparison algorithm, and constructing a multi-modal testing and evaluating data set; deep features of all modes are extracted through a mode exclusive encoder, and a unified multi-mode function vector is output based on information entropy dynamic weighted fusion; constructing triple training data, and optimizing the multi-modal large model by combining a joint loss function with an enhanced training strategy; and finally, deploying the model at a cloud end and an edge node, carrying out real-time triggering test by docking a CI / CD process, and realizing functional integrity verification, interactive compliance detection and accurate defect positioning through semantic vector comparison, graph isomorphism rate calculation and cross-modal attention backtracking. According to the scheme, the test comprehensiveness and precision are improved, and the rapid iteration requirement of software is met.

The invention relates to the field of computer technology and medical image processing, in particular to a medical image segmentation method fusing random region cutting enhancement and a pseudo label semi-supervised mechanism. In order to solve the problems of scarcity of annotation data, weak model generalization ability, inaccurate segmentation boundary and the like in a current medical image segmentation task, the invention provides a medical image segmentation method fusing random region cutting enhancement and a pseudo label semi-supervised mechanism, which is called an RCDE segmentation model. The model adopts a shared encoder and double decoder structure, combines a structure-level disturbance generation strategy, enhances the perception ability of the model for image structure change by mixing and recombining labeled images and unlabeled images, generates pseudo labels by utilizing a teacher network, and introduces a dynamic confidence coefficient screening mechanism, so that low-quality pseudo labels are effectively eliminated, and the robustness of the model is improved. The training stability and the pseudo-supervision effect are improved, preprocessing such as size normalization and image enhancement is carried out on the image, and the consistency and robustness of model input are improved.

The invention discloses a CT image analysis method and system based on a neural network, and relates to the technical field of CT image analys.The method comprises the steps that an original CT image is obtained after user authorization, a Laplace operator is adopted to strengthen a focus boundary, and a circular region of interest is intercepted to remove edge sensitive information; extracting edge and texture information in the standardized image; focus area features are focused step by step; executing characteristic distillation balance based on category sample distribution, and outputting a focus characteristic graph with local perception enhancement and sample balance characteristics; segmenting the lesion feature map into serialized units, embedding position codes, inputting the serialized units into a plurality of layers of encoders, and fusing an image structure and text indication information through a dynamic adjustment mechanism; performing linear classification on the global semantic vector to output a diagnosis result, generating a focus thermodynamic diagram, and superposing the focus thermodynamic diagram to an original image for visualization; and performing dynamic optimization based on doctor feedback. The accuracy of feature analysis is improved; the overall operation efficiency of the system is improved.

The invention provides a method and a system for restoring a bamboo-strip character image based on multi-granularity feature guidance, and innovatively designs a coarse-fine two-stage restoration network and end-to-end multi-task loss joint training aiming at the problems of structure-texture confusion, non-uniform degradation, low contrast ratio and the like of the bamboo-strip image. In the coarse repair stage, a font texture and structure double-reconstruction sub-network is used for separating semantics from a source; in the fine repairing stage, multi-scale dynamic range distribution diagram self-attention (Mdma) is provided, pixels are dynamically classified according to degradation intensity, and long-short range dependence joint modeling is achieved; an adaptive mask is designed to sense pixel shuffling downsampling (Ampd), sampling is guided by mask confidence, damage position information is kept, and artifacts are inhibited. Five mainstream methods are compared on a homemade 313 bamboo strip single word data set, the PSNR, the SSIM and the FID are optimal under 0-60% irregular masks, visual evaluation of real missing samples is natural in texture, the structure is complete, and the readability of the bamboo strip characters and the subsequent recognition accuracy are effectively improved.

The invention discloses a fusion graph recognition and symbolic expression-based resin-based composite material automatic native modeling and reverse design method and system, belongs to the field of composite material modeling and design, and particularly relates to a graph neural network and symbolic regression fusion-based composite material native modeling and structure reverse optimization method. The method comprises six steps of microcosmic image structure extraction, topological graph construction and graph embedding, constitutive relation symbol modeling, graph structure homogenization, performance-oriented reverse design and multi-modal performance prediction, and can realize an automatic process from microcosmic image to macroscopic performance prediction to structure optimization design. The problems that an existing method is low in efficiency, poor in interpretability and difficult in reverse design are solved, and the modeling efficiency and the design intelligence level of the composite material under multiple scales and multiple targets are improved.

The invention discloses a steel ladle hoisting safety monitoring and early warning method, system and equipment and a medium. The method comprises the following steps: acquiring double-side images in a steel ladle hoisting process; inputting the bilateral images into a multi-target detection model, and outputting bounding boxes, category labels and confidence coefficients of the plurality of key targets; obtaining a local cutting image of each key target according to the bilateral images and the bounding box, the category label and the confidence of each key target, and generating a structured text in combination with the bounding box position, the relative pixel distance and the category label of each key target; inputting the local cutting image, the structured text and the prompt word template of each key target into a multi-modal big language model, and outputting a behavior label, a risk level, a reason, a recommendation action and an instruction state of the current steel ladle hoisting scene through multi-modal reasoning; early warning is conducted according to the recommended action, and the steel ladle hoisting equipment is controlled according to the instruction state. According to the invention, intelligent monitoring and linkage early warning in the steel ladle hoisting process can be realized.

The invention discloses a multi-modal information fused high-myopia retina image classification method and system. The method comprises the following steps: acquiring an OCT image of a to-be-classified high-myopia retina and structured numerical data corresponding to the image; the OCT image is preprocessed, the preprocessed image is input into an image encoder for high-dimensional semantic representation learning, and overall semantic vector representation of the image is obtained; converting the structured numerical data into a medical language description text, and inputting the medical language description text into a text encoder for deep semantic modeling to obtain overall semantic vector representation of the text; inputting to a multi-modal fusion module, carrying out feature interaction and fusion through a bidirectional cross attention mechanism, and generating a fused multi-modal feature; and outputting a retina splitting stage category corresponding to the OCT image through a classification module. According to the method, the OCT image, the structured numerical data and the split staging definition text are utilized to perform multi-modal feature fusion, so that the accuracy of image staging recognition is improved.

The invention discloses a single image defogging method based on block-by-block nonlinear brightness prior, which belongs to the technical field of image processing, and comprises the following steps of: dividing a fog-containing image into local blocks, calculating the average brightness of the blocks, and constructing prior block-by-block monotone increasing nonlinear mapping to represent the corresponding relationship between the brightness of fog-containing blocks and the brightness of clear blocks; an atmospheric scattering model is combined, an atmospheric light vector is modeled into a vector, the vector and a PPWF form a parameterized recovery model, three scalar parameters are taken as a core, an optimal parameter is obtained through multi-target joint optimization, alternate optimization and golden section search, and finally a defogged image is generated. The method has the advantages of few parameters, low complexity and no need of training data, the definition and global contrast of far and near scenery can be remarkably improved while the image structure is maintained, and halo, excessive enhancement and color cast are effectively inhibited; the method has good robustness for different fog densities and illumination conditions, and is suitable for real-time and embedded defogging application of single-channel or multi-channel images.

The invention discloses a visual-text intelligent agent-based generative cross-modal retrieval method and system, and the method comprises the steps: obtaining image features, and constructing an image structured identifier; performing fine tuning on the multi-modal large language model, learning a semantic relationship between the image features and the structured identifiers, and learning a semantic relationship between the structured identifiers and query input of the user; generating a structured identifier based on the retrieval capability of the multi-modal large language model; based on the ability of the multi-modal large language model to understand multi-modal information, the most matched image is screened from the multiple images. According to the method, the two-stage query workflow is automatically constructed according to the query text input by the user, the same image identifier is allowed to correspond to a plurality of images in the coarse-grained retrieval stage, and the image which is most matched with the input of the user is screened from the plurality of images obtained by coarse-grained retrieval in the fine-grained retrieval stage.

The invention relates to the technical field of point location labeling, in particular to a self-adaptive load balancing point location labeling method and system based on image processing, and the method comprises the following steps: obtaining an image source division region, extracting an edge pixel number texture direction number color channel standard deviation, generating region detail redundant structure distribution information, and monitoring node state data. The method comprises the following steps: constructing a node resource load state mapping set, matching according to a regional detail redundancy ratio and a node state, establishing a task allocation relationship, extracting a boundary gray derivative, screening a boundary stable candidate point location set, calling point location coordinates, sending the point location coordinates to corresponding nodes, executing target verification, writing in an image frame, and generating a labeling result. Quantitative extraction is performed on distribution among image region edge pixels, texture directions and color channels, and task assignment is accurately paired in combination with image structure complexity and node states, so that resource mismatching and processing retardation are avoided, and task distribution accuracy is improved.

The invention relates to the technical field of remote sensing image processing, in particular to an image super-resolution system and method based on high and low frequency separation perception Mama, and the method comprises the steps: firstly carrying out the shallow convolution feature extraction of a low-resolution image; then entering a plurality of frequency sensing Mama groups, performing frequency separation and enhancement on each group through a high and low frequency feature adaptive enhancement module, and performing depth feature transformation through a plurality of frequency sensing Mama blocks; the extracted depth features are refined through a global channel-space attention module, and finally a high-resolution image is reconstructed through up-sampling. Through organic combination of the modules, the defects of insufficient frequency perception, low global modeling efficiency, insufficient feature optimization and the like are effectively overcome, and high-quality collaborative reconstruction of remote sensing image structures and textures is realized.

The invention discloses an unmanned aerial vehicle fine inspection method and system for a beam-pumping unit. An unmanned aerial vehicle automatically flies according to a preset route to obtain an image of the pumping unit and performs defect identification. Constructing and deploying a multi-point fine inspection model based on deep learning, and automatically identifying, positioning and splitting component sub-images of eighteen standard inspection points from integral images shot at different angles and distances; respectively inputting the sub-images into one-to-one corresponding special small model defect detectors for parallel analysis to obtain a preliminary result containing defect categories and bounding box coordinates; and sending the defect judgment of which the confidence coefficient is lower than a threshold value or the data sparse category and the corresponding sub-image into a multi-modal visual large model for secondary semantic level reinspection. Through a three-stage collaborative analysis process, structural decomposition of a complete machine image is realized, a special small model efficiently screens known defects, a large model improves the fuzzy, rare or complex defect recognition capability, and the defect recognition automation level and the judgment precision are remarkably improved.

The invention discloses a defect detection method and system based on insulator multi-modal image fusion, and relates to the technical field of electrical equipment defect detection, multi-modal image fusion is performed based on an attention mechanism improved RFN-Nest image fusion model, fusion of infrared image temperature anomaly features and visual image structure detail features is enhanced, and the defect detection accuracy is improved. The information entropy, mutual information and other indexes of the generated fusion image are remarkably superior to those of an original model and a traditional fusion method, high-quality data support is provided for a detection task, then defect detection is carried out based on an attention mechanism improved YOLOv8 target detection model, the defect feature discrimination capability and the anti-interference capability are improved, and the detection efficiency is improved. The problems of missing report and false report of insulator defects in a complex scene are effectively solved, the performance is remarkably improved compared with a traditional independent link design scheme, and the method can be directly applied to an actual electric power inspection scene.

The invention discloses a structure safety assessment method and system based on a robot holder image, and relates to the field of structure safety assessment. The method comprises the following steps: S1, controlling a robot carrying a holder and a multi-modal sensor group to collect a surface image, internal heat distribution and three-dimensional contour data of a target structure, constructing a model based on an improved deep learning network to identify and classify surface defects and internal hidden damages, quantifying the defects through an algorithm, and outputting the data; s2, constructing a digital twin based on the three-dimensional contour data, mapping design parameters, tracking a dynamic feature point displacement trajectory to calculate deformation parameters, and outputting a deformation damage analysis result through a correlation model; and S3, calling a pre-constructed reinforcement learning adaptive evaluation model, inputting defect quantitative data and a deformation damage analysis result, and outputting a safety level and a risk evaluation result combined with a preset specification. The method improves the evaluation precision and efficiency, and adapts to the evaluation requirements of different target structures.

The invention relates to the technical field of three-dimensional modeling, in particular to a three-dimensional reconstruction method and device for a two-dimensional drawing, equipment and a medium. The method comprises the steps of firstly obtaining to-be-reconstructed two-dimensional drawings of at least two views in target object engineering views and collecting features to obtain feature points and geometric information, then constructing a target graph structure, updating nodes based on a graph neural network and clustering to obtain an overall view, converting the overall view into view feature vectors, and constructing a target image structure; a three-dimensional contour is calculated through an inference model containing a multi-view self-attention mechanism and an alignment model introducing projection consistency geometric constraints; and finally, three-dimensional space data is extracted based on a symbol distance function, and shape processing is performed to construct a three-dimensional grid model. Therefore, the two-dimensional drawing is automatically and accurately subjected to view recognition and three-dimensional alignment, and a three-dimensional grid model is efficiently completed.

The invention discloses a defect detection sample data generation method and related equipment, and relates to the technical field of defect detection sample data generation, and the defect detection sample data generation method comprises the steps: obtaining a good product background image and an expected defect direction; based on the expected defect direction, respectively generating a structure guide mask and a semantic guide text cue word according to the defect knowledge base; inputting the non-defective product background image, the structure guiding mask and the semantic guiding text cue word into a local redrawing model, synthesizing a local defect image, and determining defect labeling information of the local defect image according to the structure guiding mask; and integrating the local defect image and the defect labeling information, and outputting defect detection sample data. According to the method, the structure guide mask and the semantic guide text cue word are generated through the defect knowledge base and are input into the local redrawing model to synthesize the local defect image, so that the defect image is effectively generated, and finally, the beneficial effect of improving the accuracy when the defect sample image is used for defect detection is achieved.

The invention relates to the technical field of intelligent detection and automatic control, in particular to a coal moisture content intelligent sensing and dust suppression control method based on hyperspectrum, which comprises the following steps: acquiring an image and sorting a gray interval, removing a high-anti-interference area, segmenting numbered blocks, counting a gray direction reversal identification interference area, extracting peak positions of each wave band, and calculating difference value groups. And aggregating the continuous image blocks to extract gray boundaries, matching the boundaries and spraying serial numbers to adjust and control, and outputting the moisture content and a dust suppression image. According to the method, a region segmentation method driven by reflection density is introduced, gray level warping and interference contour elimination are combined, a block set is constructed to enhance the processing pertinence, through multi-band peak position difference value and symbol sequence continuity judgment, peak position change grouping is established to identify the spectrum trend, boundary aggregation is completed in the gray level mean value difference value direction, and finally, the spectrum trend is identified. And a control chart is generated by combining serial number position information, and linkage of an image structure, spectral distribution and control feedback is realized.

The invention discloses a visual displacement estimation method combining super-resolution and Kalman filtering, and belongs to the technical field of structural health monitoring. The method comprises the following steps: extracting an image frame by frame from an original video, generating a high-resolution image by adopting a super-resolution generative adversarial network (ESRGAN), and calculating a scale factor; respectively acquiring the pixel displacement and the pixel speed of the high-resolution image structure by using a UCC template matching method and a KLT optical flow method, and converting the pixel displacement and the pixel speed into real displacement and real speed through scale factors; and carrying out data fusion on the real displacement and the real speed by adopting adaptive Kalman filtering (AKF) to obtain the accurate displacement of the structure. According to the method, the image quality is improved through the super-resolution, the multi-source information fusion technology is combined, the problems of image resolution limitation, noise interference and insufficient robustness in vision measurement are effectively solved, a high-precision and low-noise displacement estimation result can be obtained, and the method is suitable for engineering scenes such as structural health monitoring.

The invention provides a semantic guidance-based container image enhancement method and system under a complex background. The method is applied to the technical field of data enhancement, a residual connection edge enhancement module and a dynamic residual semantic matching module are introduced, a loss function module is designed, and high-fidelity enhancement of container image structure details and color features in a complex environment is realized; according to the invention, the residual connection edge enhancement module explicitly extracts and adaptively enhances edge features through local edge enhancement, enhances detail structures and edge texture features in an image, and effectively alleviates the problem of detail degradation in a deep network; according to the dynamic residual semantic matching module, directional guidance and activation of high-level semantic information on shallow-layer features are realized, and the sensing and positioning capability of the model on a container target area is effectively enhanced.

The invention discloses a non-reference image quality evaluation method and system based on a multi-scale image structure, and the method comprises the steps: firstly, for an input image, employing a multi-branch convolutional neural network, and respectively extracting multi-dimensional quality feature representations; secondly, on the basis of multi-dimensional quality feature representation, a quality relation graph is constructed on different spatial granularities, and global pooling is carried out respectively to obtain global quality representation on different spatial granularities; and then constructing a graph attention fusion module, carrying out weighted fusion on the global quality representations on different spatial granularities, and generating a comprehensive quality representation. Finally, the comprehensive quality features are sent into a multi-layer perceptron regression head, and a final image quality score is output. And training is carried out. According to the invention, the one-sidedness of single-dimensional evaluation is avoided, the quality evaluation is more comprehensive and accurate, and the accuracy and generalization ability of no-reference image quality evaluation are improved.

The invention provides a dynamic PET image and kinetic parameter prediction method and device and a storage medium. The method comprises the steps that first N frames of dynamic PET image data are obtained and preprocessed to obtain a standard image sequence; and inputting the standard image sequence into the prediction model to obtain subsequent frame dynamic PET image data and a corresponding kinetic parameter map. The prediction model comprises an image structure sensing channel for extracting image structure features, a dynamic modeling channel for extracting dynamic parameter features, a cross fusion module for fusing the image structure features and the dynamic parameter features, and a prediction reconstruction module for performing prediction reconstruction based on a fusion result. The method has the advantages that image structure features and kinetic parameter features can be fused on the basis of the dynamic PET image early-stage frames, scanning time and radiation dose are reduced, and meanwhile follow-up frame dynamic PET image data with high image quality and dynamic interpretability and the corresponding kinetic parameter atlas are generated.

The invention discloses an unmanned aerial vehicle small sample target recognition method based on spatial information mining, and aims to solve the problems that an existing small sample method does not fully utilize an image structure in a scene, ignores a semantic relationship between support and query samples and is limited in recognition precision. The core is a query-support transformer framework for unified modeling of global and local structural relationships of samples, the framework comprises a cross-scale interactive feature extractor, a sample transformer and a block transformer module, a task-type mechanism is adopted in the training stage, enhanced features are generated by images through the extractor, the sample transformer constructs a global semantic relationship, and the overall semantic relationship is obtained. The method comprises the following steps: realizing local block-level matching by a block transformer, performing weighted summation on similarity of the two to obtain a matching score, combining weighted sum of comparison loss and cross entropy loss, updating parameters by using momentum stochastic gradient descent, constructing a task based on a trained model in a test stage, extracting features, and fusing global and local similarity to complete classification. And the unmanned aerial vehicle image small sample classification performance is effectively improved.

The invention relates to the technical field of image recognition, and discloses an image feature recognition method and system applied to the early stage of kidney diseases. According to the method, the features in the original ultrasonic image are extracted in parallel, and the micro texture change and the macroscopic form information shown in the abnormal area can be captured respectively, so that the limitation of a traditional single-scale feature extraction method in coping with diffusive and weak-saliency image features is overcome; the local texture feature map and the global structure feature map are analyzed, quantization and positioning of weak texture changes are achieved, priori knowledge is converted into objective indexes, pixel-level fusion is carried out on a texture abnormal map and a region importance weight map, and through the knowledge-guided and data-driven deep fusion method, the deep fusion of the texture abnormal map and the region importance weight map is achieved. It is ensured that the finally recognized abnormal area is not only a statistical outlier but also conforms to importance distribution of an image structure, and high-precision automatic recognition of weak and diffuse features in the kidney medical image is facilitated.

The invention discloses a multi-scale infrared polarization image enhancement method based on MSD-PCNN, and relates to the technical field of infrared image processing, and the method comprises the steps: firstly, carrying out the multi-scale and multi-direction decomposition of an input intensity image I and a polarization feature map DoLP based on NSST, and obtaining a low-frequency sub-band and a high-frequency sub-band; secondly, according to the obtained low-frequency sub-band, designing a low-frequency sub-band fusion rule based on distance difference perception; and finally, according to the obtained high-frequency sub-band, designing a high-frequency sub-band fusion rule based on MSD-PCNN, and fully extracting texture and edge detail features. According to the MSD-PCNN-based multi-scale infrared polarization image enhancement method provided by the invention, redundant information interference is suppressed, the hierarchical definition of an image structure is improved, and meanwhile, the visual comfort and global naturalness of a fused image are also improved.

The invention discloses an unsupervised infrared image enhancement method based on frequency dynamic fusion and perception optimization, and the method comprises the steps: constructing an infrared enhancement diffusion model at a back diffusion stage of a diffusion model, the infrared enhanced diffusion model sequentially comprises an infrared image feature fine tuning stage, a frequency dynamic mask fusion guide stage and a CDF perception optimization stage according to an input and output sequence. According to the method provided by the invention, effective contrast enhancement is carried out while overall brightness improvement is carried out, an original image structure is reserved, details are enhanced, the details and structure information can be fully reserved when the image is generated by adopting the method, the problems of overexposure and uneven brightness are effectively inhibited, and the image quality is improved. According to the method, the output result shows more natural and harmonious characteristics on image quality and brightness levels, has good cross-dataset generalization ability, also has a good enhancement effect on other infrared datasets of non-active thermal excitation, and effectively improves the infrared image quality.

The invention belongs to the technical field of image quality evaluation and computer vision, and particularly relates to a super-resolution image quality evaluation method based on a structure-texture separation and dynamic perception mechanism, which comprises the following steps of: firstly, acquiring an image quality evaluation data set of a training image set containing super-resolution and corresponding low-resolution images; generating an image data set; then obtaining a super-resolution image quality evaluation model based on a structure-texture separation and dynamic perception mechanism, wherein the super-resolution image quality evaluation model comprises a structure and texture decomposition module, a self-adaptive image feature extraction module, a dynamic texture guide module, a texture guide block, a gating dynamic interaction block and a mixed-order global pooling module; on the basis, image structures and textures are explicitly separated in the feature extraction stage, structure branches and texture branches are established respectively for independent modeling, and the defect that structures and textures are processed in a unified mode through an existing method is overcome.

Disclosed is a joint denoising and demosaicking method for a color RAW image guided by a monochrome image. The method comprises: constructing a synthetic image dataset of a monochrome-color binocular camera system for training and testing of network modules; constructing an aligned guidance image generation module by utilizing a structural correlation between a monochrome image and a color image, using a clean grayscale image corresponding to the RAW image as supervision, and training the aligned guidance image generation module by using a perceptual loss function to generate a high-quality aligned guidance image; using the generated aligned guidance image to guide the joint denoising and demosaicking process of the color RAW image; and training a guided denoising and demosaicking module ensure the accuracy of the color of the denoising and demosaicking result while accurately transferring the guidance image structure.

The invention discloses a focal length dynamic regulation and control method and system for target secondary confirmation, and relates to the technical field of computer vision and image processing. The focal length dynamic regulation and control method and system for the secondary confirmation of the target comprises the following steps: S1, taking over a real-time video stream of a dome camera, and collecting and preprocessing image structure basic data; s2, based on the image structure basic data, target attributes are recognized, and state fluctuation is judged; s3, by taking the focusing mark as a control basis, calculating a focusing magnification by combining spatial characteristics and state fluctuation; and S4, constructing an initial abnormal behavior sample library, and executing similarity evaluation of abnormal structures in the abnormal behavior sample library based on the behavior feature sequence extracted from the focusing image. The problems that secondary confirmation cannot be carried out and related information cannot be preserved when the suspicious target recognition confidence coefficient is insufficient are solved.

According to the PDF document structuring processing and conversion tool, contents (including texts, tables, images, structures and the like) in PDF documents are analyzed and converted into structured data (Markdown) capable of being used for downstream LLM tasks, and the PDF structuring processing and conversion tool is particularly suitable for PDFs of complex typesetting of academic literatures, reports and the like. Through a literature vectorization function, medical professional literatures can be precisely processed, core ideas, key data and logic relations in the literatures can be deeply understood, vectors capable of accurately reflecting the essence of the literatures can be generated, important information is prevented from being lost or misread, and information processing and application effects are improved. Meanwhile, the personalized Agent can perform adaptive optimization aiming at diversified tasks in the medical academic field, and can fit research directions, knowledge backgrounds and working habits of different users, for example, the personalized Agent provides support conforming to a rigorous logic structure and standard academic expression during academic paper writing; and when the academic slide is manufactured, concise and clear contents with highlighted key points can be presented, so that the working efficiency of medical workers and researchers is effectively improved, and the high-standard requirements of the medical field are met.

The invention provides a response duration analysis method and device, a program product and electronic equipment, and relates to the technical field of computers. The method comprises the steps of recording a target video in response to a target application program starting operation; performing video analysis processing on the target video to obtain an image frame sequence; setting a starting reference image and an ending reference image according to the image frame sequence; determining a starting frame image according to the structural similarity value of the frame image in the image frame sequence and the starting reference image; according to the structural similarity value of the frame image in the image frame sequence and the ending reference image, determining an ending frame image; and determining the response duration of the target application program according to the frame number between the start frame image and the end frame image. According to the method, the starting point and the ending point are automatically positioned in the automatically recorded target video by using the image structure information, so that the response duration of the target application program is accurately determined.