Ski field outdoor advertisement monitoring data analysis method and system

Abstract

The invention relates to the technical field of outdoor advertisement effect evaluation, in particular to a ski field outdoor advertisement monitoring data analysis method and system.The method comprises the steps that multi-source monitoring data of ski field outdoor advertisement positions are collected, and the multi-source monitoring data comprise advertisement position video streams, environment sensing data and skier behavior tracks; based on the environment sensing data, executing dynamic environment interference correction on the advertisement position video stream, and generating an advertisement exposure image sequence; according to the advertisement exposure image sequence and the skier behavior track, constructing a dynamic space-time correlation between the skier and an advertisement position, and extracting an advertisement attention feature set; and inputting the advertisement attention feature set into an advertisement effect evaluation model to generate an advertisement effect quantitative index. According to the invention, the problem of advertisement attention quantification misalignment in the high-speed motion scene of the ski field is effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of outdoor advertising effect evaluation, and in particular to a method and system for analyzing outdoor advertising monitoring data at a ski resort. Background Art

[0002] Outdoor advertising monitoring data analysis technology aims to evaluate the effectiveness of advertising exposure by collecting video, sensor and user behavior data from advertising locations. In ski resort scenarios, it needs to cope with the unique challenges of skiers' high-speed movement and complex environmental interference. Its core is to convert multi-source data into quantitative indicators of advertising effectiveness to optimize delivery strategies.

[0003] Current mainstream solutions rely on wide-angle camera video analysis or single-positioning technology, some with the incorporation of environmental sensors. However, these technologies suffer from fundamental flaws: The quantification of ad attention in high-speed skiing scenarios is severely inaccurate. Specifically, wide-angle camera solutions produce motion blur when skiers are traveling at high speeds. Environmental changes cause image distortion, making traditional hardware filters or linear corrections unable to recover significant details, resulting in computational distortion. Single-positioning technology suffers from high signal loss rates in complex ski resort terrain, making it impossible to continuously track behavior, leading to failures in extracting features such as ad dwell frequency and trajectory deviation. These flaws directly point to a core issue that existing technologies cannot address: the distortion of advertising effectiveness evaluation caused by the coupling of high-speed motion and environmental interference.

[0004] The information disclosed in this background technology section is only intended to deepen the understanding of the overall background technology of the present disclosure and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art known to those skilled in the art. Summary of the Invention

[0005] The present invention provides a method and system for analyzing outdoor advertising monitoring data in a ski resort, which can effectively solve the problems in the background technology.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is: A method for analyzing monitoring data of outdoor advertising at a ski resort, the method comprising: Collecting multi-source monitoring data of outdoor advertising spaces at ski resorts, including video streams of advertising spaces, environmental sensor data, and skier behavior trajectories; Based on the environmental sensing data, dynamic environmental interference correction is performed on the video stream of the advertising position to generate an advertising exposure image sequence; Based on the advertisement exposure image sequence and the skier's behavior trajectory, a dynamic spatiotemporal association between the skier and the advertisement position is constructed, and an advertisement attention feature set is extracted; The advertisement attention feature set is input into an advertisement effect evaluation model to generate an advertisement effect quantitative index.

[0007] Furthermore, generating an advertisement exposure image sequence includes: Acquiring the environmental sensor data, including light intensity, snow and fog concentration, and wind speed data; Constructing a brightness compensation matrix according to the light intensity to correct the advertising position video stream; Compensating for motion blur of the video stream of the advertising spot based on the wind speed data; A defogging convolution kernel is generated based on the snow fog concentration, and the advertising position video stream is corrected according to the defogging convolution kernel to generate the advertising exposure image sequence.

[0008] Furthermore, the generation of the skier's behavior trajectory includes: Get the skier's starting point on the trail and generate the initial position coordinates; Based on the real-time changes of the initial position coordinates, the skier's clothing color and ski equipment outline features corresponding to the moving area are obtained and combined to generate dynamic visual features; The skier's behavior trajectory is generated by combining the initial position coordinates with the dynamic visual features.

[0009] Furthermore, dynamic visual features are generated by merging, including: Performing spatial transformation on the color features of the skier's clothing to extract hue and saturation components to generate a first feature vector; Performing key point detection on the ski equipment contour features, obtaining a contour curvature histogram and generating a second eigenvector; combining the first feature vector and the second feature vector of each skier to output the dynamic visual feature; An attention weight is added to the combination of the first feature vector and the second feature vector, and the attention weight is dynamically adjusted according to the real-time distance between the skier and the advertising space.

[0010] Furthermore, the dynamic spatiotemporal association includes: Taking the visible area of ​​the advertisement exposure image sequence as a reference plane, projecting the skier's behavior trajectory onto the reference plane; When the skier's behavior trajectory enters the reference surface, the gaze duration timer is started; When the angle between the skier's behavior trajectory and the normal direction of the reference surface and the gaze duration are less than a threshold, it is recorded as valid gaze data.

[0011] Furthermore, the advertisement attention feature set is generated by aggregating the effective gaze data, wherein an attention weight is added during the aggregation process as an accompanying weight of the effective gaze data.

[0012] Furthermore, quantitative indicators of advertising effectiveness are generated, including: After the effective gaze data in the advertisement attention feature set is converted into the frequency domain, the probability of behavioral conversion guided by the advertisement is predicted by combining the attention weight and historical behavior data, and the quantitative index of the advertisement effect is output.

[0013] Furthermore, it also includes collecting real-time weather warning data, and when the weather warning data reaches a threshold, adding a confidence correction coefficient to the advertising effect quantitative index.

[0014] A ski resort outdoor advertising monitoring and data analysis system, comprising: The data acquisition module collects multi-source monitoring data of outdoor advertising spaces at ski resorts, including video streams of advertising spaces, environmental sensor data, and skier behavior trajectories. The delayed image module performs dynamic environmental interference correction on the video stream of the advertising slot based on environmental sensor data to generate an advertising exposure image sequence; The feature generation module constructs the dynamic spatiotemporal association between skiers and ad spots based on the ad exposure image sequence and skier behavior trajectory, and extracts the ad attention feature set; The effect quantification module inputs the advertising attention feature set into the advertising effect evaluation model to generate advertising effect quantitative indicators.

[0015] Furthermore, the feature generation module includes: A reference setting unit, which takes the visible area of ​​the advertisement exposure image sequence as a reference plane and projects the skier's behavior trajectory onto the reference plane; The gaze timing unit starts the gaze duration timing when the skier's trajectory enters the reference surface; The data recording unit records valid gaze data when the angle between the skier's behavior trajectory and the normal direction of the reference surface and the gaze duration are less than a threshold.

[0016] The technical solution of the present invention can achieve the following technical effects: By constructing multimodal data fusion in the high-speed motion scene of the ski resort, an accurate quantitative assessment of advertising attention is achieved, effectively overcoming the inaccurate quantification effect problems caused by dynamic blur, environmental interference and trajectory breakage.

[0017] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 The present invention is a flowchart of a method for analyzing outdoor advertising monitoring data at a ski resort; Figure 2 A schematic diagram of the process for generating an advertisement exposure image sequence; Figure 3 Schematic diagram of the process for obtaining skier behavior trajectory; Figure 4 A conceptual diagram of dynamic spatiotemporal association. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Example 1

[0022] like Figure 1 As shown, the present application provides a method for analyzing outdoor advertising monitoring data at a ski resort, the method comprising: S10: Collect multi-source monitoring data of outdoor advertising spaces at ski resorts, including video streams of advertising spaces, environmental sensor data, and skier behavior trajectories; S20: Based on the environmental sensing data, dynamic environmental interference correction is performed on the video stream of the advertising position to generate an advertising exposure image sequence; S30: Based on the advertisement exposure image sequence and the skier behavior trajectory data, a dynamic spatiotemporal association between the skier and the advertisement position is constructed, and a feature set of advertisement attention is extracted; S40: Input the advertisement attention feature set into the advertisement effect evaluation model to generate advertisement effect quantitative indicators.

[0023] Specifically, first, a multi-source data acquisition system is set up and deployed at several key locations in the ski resort, including high-definition network cameras facing the advertising space, environmental monitoring nodes covering the site, and behavior acquisition terminals installed on skiers. The video cameras continuously collect video streams from the advertising space to ensure that all-weather image data can be obtained. The environmental sensor nodes collect environmental data in real time under the corresponding time and space, providing a basis for subsequent video correction, and the skier behavior acquisition terminal records the movement trajectory data of the skier in the site, including information such as position, speed, dwell time, and direction of travel. In the data processing stage, first, according to the environmental sensor data, the system collects the data from the skier's body, so as to obtain the image data of the skier. The video image is corrected for dynamic environmental interference. Specifically, when the sensor detects interference conditions such as environmental factors such as strong light reflection, snow and fog obstruction, backlight or low light at night, the video image is corrected through image processing methods such as image enhancement, regional occlusion analysis and background modeling to remove or weaken the image distortion caused by the above interference, thereby obtaining a higher quality advertising exposure image sequence; then, a dynamic spatiotemporal association is constructed between the advertising exposure image sequence and the skier trajectory data. To achieve this association, the skier's trajectory is synchronized in time and spatially mapped, and its position at a specific moment is cross-judged with the visible area of ​​the advertising position. If the skier's movement direction at a certain moment is towards the advertising space and the distance is within the visual perception range, the skier is considered to be in a potential attention state; if the skier's head is further detected to be facing the advertising direction in the image, the state is upgraded to an effective attention state. In this way, every effective interaction between the skier and the advertising space can be marked and counted; then, based on the constructed spatiotemporal correlation relationship, the advertising attention feature set is extracted. The feature set includes but is not limited to: the number of people paying attention per unit time, the average attention time, the distribution of attention angles, the distribution of stay time, the advertising viewability rate and other dimensions, in order to improve the effectiveness of the feature set. The data can be classified and analyzed based on the skier behavior type to screen out behavioral samples that have practical significance for advertising effectiveness. Finally, the extracted advertising attention feature set is input into a pre-trained advertising effectiveness evaluation model. The model can be constructed based on machine learning, such as using random forests, support vector machines or neural networks, and supervised training through historical advertising data to output comprehensive evaluation indicators, such as advertising exposure value points, attention heat levels or advertising effectiveness scores. Preferably, the model can be dynamically updated and adjusted according to actual operational needs to adapt to the effectiveness evaluation of different advertising delivery strategies.

[0024] Through the technical solution of the present invention, multimodal data fusion is constructed in the high-speed motion scene of the ski resort, which realizes the accurate quantitative evaluation of advertising attention and effectively overcomes the problem of inaccurate quantification effect caused by dynamic blur, environmental interference and trajectory breakage.

[0025] Further, if Figure 2As shown, an advertisement exposure image sequence is generated, including: Acquire environmental sensor data, including light intensity, snow and fog concentration, and wind speed data; Construct a brightness compensation matrix based on the light intensity to correct the video stream of the ad slot; Compensate for motion blur in the ad slot video stream based on wind speed data; A defogging convolution kernel is generated based on the snow fog concentration. The video stream of the advertising position is corrected according to the defogging convolution kernel and an advertising exposure image sequence is generated.

[0026] As a preferred embodiment of the above, first, the light intensity data of the current time period is collected from the environmental sensor at regular intervals, and a set of brightness compensation matrices is constructed based on the data to perform brightness correction processing on the collected advertising position video stream. The construction process of the brightness compensation matrix comprehensively considers the overall brightness level of the video image, the distribution characteristics of the local over-exposed or under-exposed areas, and the periodic pattern of light changes over time, so as to achieve dynamic balanced correction of the picture brightness. Especially in the strong light reflection or shadow occlusion scenes commonly seen in ski resorts, this brightness compensation mechanism can effectively improve the visual clarity of the advertising area and ensure the stability of the image quality relied on by subsequent analysis; secondly, wind speed is an important external factor in the generation of dynamic blur. In this embodiment, The embodiment is used to perform dynamic blur compensation for the video stream of the advertising position. By obtaining the wind speed data of the video acquisition area in real time, combining the camera acquisition parameters and the motion blur characteristics of the target edge in the video image, the blur caused by the wind speed in the image, such as slight shaking, screen dragging or target ghosting, is judged. In order to compensate, the adaptive deconvolution method is used in conjunction with the image sharpening strategy. The image edge and texture structure are restored through the blur estimation model to enhance the boundary contrast and detail restoration effect of the advertising area. Preferably, multiple wind speed threshold intervals can be set, and different intervals correspond to different degrees of blur models, thereby improving the pertinence and accuracy of image restoration; further, for the obvious effect of snow fog concentration on image quality, In view of the significant impact, this embodiment introduces a defogging convolution kernel generated based on snow fog concentration to perform defogging on video images. By deploying visibility sensors near the advertising space or utilizing the camera's built-in image analysis function, the snow fog concentration index in the current picture is calculated in real time. When thick fog or snowflakes are detected with obvious obstruction, the defogging convolution kernel matching the current snow fog level is automatically loaded. This convolution kernel is designed based on the principle of image defogging and can suppress the brightness diffusion and detail loss problems caused by snow fog in the image processing stage. In order to enhance the defogging effect, an image contrast enhancement mechanism and a color reconstruction algorithm are also introduced to further restore the clarity and color saturation of the advertising image on the basis of spatial domain filtering. In the integration After the above processing steps, the corrected image frames are combined in chronological order to generate a set of stable, clear, and minimally affected advertising exposure image sequences. This image sequence serves as an important basic dataset for subsequent skier behavior analysis and advertising attention extraction. It not only effectively solves the adverse effects of complex weather environments on video quality, but also provides high-quality image-level support for the evaluation of outdoor advertising effects in ski resorts. For example, in a certain period of heavy snow, by accurately identifying the snow fog concentration and constructing the defogging convolution kernel in real time, the image clarity was greatly improved without replacing the hardware equipment. This enables the subsequent advertising recognition module to still accurately determine the location and content of the advertisement, thereby ensuring the continuity and effectiveness of data analysis.

[0027] Further, if Figure 3As shown in FIG, the generation of skier behavior trajectory data includes: Get the skier's starting point on the trail and generate the initial position coordinates; Based on the real-time changes of the initial position coordinates, the skier's clothing color and ski equipment outline features in the corresponding moving area are obtained and merged to generate dynamic visual features; The initial position coordinates and dynamic visual features are combined to generate skier behavior trajectory data.

[0028] As a preferred embodiment of the above, first, the initial position coordinates of the skier are obtained by setting a positioning sensor near the starting point of the ski slope. When the skier is ready to set off, the sensor will capture the geographic location coordinates at the time of departure and record them as the initial coordinates. The initial position coordinates will serve as the basis for subsequent trajectory calculation and dynamic behavior tracking. After obtaining the initial position coordinates, the real-time movement state and surrounding environment of the skier are further monitored, and the movement area of ​​the skier is analyzed based on the changes in the initial coordinates. Specifically, combined with real-time camera image acquisition data, the skier's clothing color and ski equipment outline are identified through target detection and image segmentation algorithms. The skier's clothing color and ski equipment shape outline are important visual features for identifying the skier's identity and dynamic behavior. Image processing techniques, such as color space conversion and contour extraction, are used to identify the skier's clothing color and ski equipment in a specific time period. These visual features, together with the initial position coordinates, constitute the skier's dynamic visual features during movement. By merging the skier's initial position coordinates with its dynamic visual features, the skier's behavioral trajectory can be accurately tracked. This process involves the collaborative work of multiple modules: First, based on the real-time changes in the initial position, the skier's The motion direction, speed, and position on the ski slope are updated. Then, the skier in each frame of the image is identified and located based on the visual features, ensuring that the trajectory data at each moment matches the corresponding visual information. The introduction of dynamic visual features makes it possible to distinguish different skiers, even if they are in the same or similar positions, and to perform independent behavioral analysis based on the color of their clothing and the outline of their ski gear. For example, in a typical ski resort application scenario, suppose skier A starts skiing from the starting point of the ski slope. The coordinates of his starting point are recorded by the initial position coordinates. As skier A moves, his motion trajectory is captured in real time by the camera deployed in front of him. Combined with the visual features of his red jacket and yellow skis, his motion trajectory is accurately calibrated. Even in complex skiing environments, the skier's clothing color and ski gear outline features can be used to eliminate background noise or interference from other skiers, thereby generating accurate behavioral trajectory data for skier A. This behavioral trajectory data not only includes the skier's location information, but also dynamic parameters such as his speed, acceleration, and motion path at different time nodes. This data can be used for subsequent advertising attention analysis and effectiveness evaluation, providing important support for the optimization of advertising delivery strategies.

[0029] Furthermore, dynamic visual features are generated by merging, including: Performing spatial transformation on the color features of the skier's clothing to extract hue and saturation components to generate a first feature vector; Perform key point detection on the ski equipment contour features, obtain the contour curvature histogram and generate the second eigenvector; Merge the first eigenvector and the second eigenvector of each skier to output dynamic visual features; Among them, the attention weight is added to the combination of the first eigenvector and the second eigenvector, and the attention weight is dynamically adjusted according to the real-time distance between the skier and the advertising space.

[0030] As a preferred embodiment of the above, first, the color features of the skier's clothing are spatially transformed by image processing technology, specifically using a color space conversion algorithm. In this process, the hue and saturation components of the clothing are extracted. The hue represents the type of color, while the saturation describes the purity or intensity of the color. By extracting these two components, the color features of the skier's clothing can be obtained. These features are crucial for tracking skiers in complex background environments. The extracted hue and saturation components are normalized to generate a first feature vector, which represents the color features of the skier's clothing; secondly, key point detection is performed on the contour features of the skier's ski equipment. In order to accurately capture the appearance features of the skier's ski equipment, the edge contours of the ski equipment are extracted by image segmentation and contour detection algorithms. Subsequently, the curvature histogram of the ski equipment is calculated based on the contour features. This is a statistical feature that can describe the change in contour shape. The curvature histogram can reflect the degree and direction of the contour, thereby helping the system to identify the differences between different ski equipment shapes. Through these operations, a second feature vector is generated, which represents the morphological features of the skier's ski equipment; next, the extracted The first and second eigenvectors of the image are merged to generate the skier's dynamic visual features. To ensure that the priorities of visual features vary across time and space, an attention weighting mechanism is introduced to dynamically adjust the combined weights of the first and second eigenvectors. These attention weights are determined by the skier's real-time distance from the ad spot. Specifically, when the skier is close to the ad spot, the weight of the clothing color features is increased to highlight the skier's visual recognition. When the skier is farther away from the ad spot, the weight of the ski equipment contour features is increased, thereby improving the recognition accuracy of the ski equipment. This process ensures that the system can flexibly respond to changes in the relative position of the skier and the ad spot in different scenarios and optimize the performance of visual features. Finally, the merged dynamic visual features are output and used as input data for subsequent modules such as advertising exposure analysis and skier behavior recognition. This feature synthesis and weighting strategy effectively improves skier recognition accuracy, especially in complex ski resort environments. It can stably track the skier's movement trajectory and behavioral characteristics, maintaining efficient and accurate data analysis at both long and close distances.

[0031] Further, if Figure 4 As shown, dynamic spatiotemporal correlation includes: Taking the visible area of ​​the advertisement exposure image sequence as the reference plane, the skier's behavior trajectory is projected onto the reference plane; When the skier's trajectory enters the reference plane, the gaze duration timer starts; When the angle between the skier's behavior trajectory and the normal of the reference surface and the gaze duration are less than the threshold, it is recorded as valid gaze data.

[0032] As a preferred embodiment of the above, first, the advertisement visible area is extracted from the advertisement exposure image sequence and defined as a planar reference plane. The reference plane is usually a two-dimensional projection area of ​​the viewing angle of the camera device facing the advertisement position, and its spatial position and orientation are determined by the camera parameters and the advertisement installation angle. By utilizing the ski resort three-dimensional environment modeling system and the spatial annotation information of the advertisement spots, the reference plane can be accurately mapped to the coordinate system of the entire ski resort; then, based on the skier's behavior trajectory data, the skier's motion path in the three-dimensional space is projected onto the above-mentioned reference plane. The projection process takes into account the skier's real-time position information and viewing angle direction, thereby ensuring the mapping accuracy of the trajectory on the reference plane; secondly, when the skier's trajectory enters the projection area of ​​the reference plane for the first time, the gaze duration timing module is automatically started. Specifically, when the skier's projection trajectory intersects or overlaps with the advertisement reference plane, the time is recorded, and the skier's position and orientation changes relative to the advertisement area are continuously tracked to determine whether he is in an attention state. The core indicators of the state include two elements: gaze duration and observation angle. The gaze duration is obtained by accumulating timestamps, recording the total time the skier maintains gaze within the projection range. The observation angle is calculated based on the angle between the skier's gaze direction and the normal vector of the advertising reference surface. The smaller the angle, the closer the skier's gaze is to the advertisement, and the higher the possibility of visual contact with the advertisement. To determine whether the skier's gaze is valid, two thresholds are set: a minimum gaze duration threshold to eliminate misjudgments caused by unintentional brief glances; and a maximum angle threshold to ensure that the gaze direction is close enough to the advertisement location. When the skier's behavior trajectory enters the advertisement reference surface, the gaze duration exceeds the set threshold and the angle between the gaze and the advertisement surface normal is lower than the angle threshold, the data is recorded as valid gaze data. This data is not only used for subsequent advertising exposure heat analysis, but can also be further combined with factors such as the skier's identity data, skiing route, and residence time to form a complete advertising influence evaluation model.In a preferred embodiment of the present invention, in order to realize the dynamic spatiotemporal correlation between the advertising space and the skier's behavior trajectory, a spatial visible area is first established as a reference plane based on the field of view of each advertising space in the advertising exposure image sequence. The reference plane can be accurately determined by multi-camera calibration technology combined with known advertising installation position information, and mapped to the unified three-dimensional coordinate system of the ski resort. The advertising reference plane not only includes two-dimensional position, but also carries normal vector direction information for subsequent judgment of the skier's observation angle. In the process of identifying the skier's real-time three-dimensional trajectory, the skier's spatial position is obtained by using a depth camera, millimeter wave radar or multi-camera image fusion, and the current line of sight direction is estimated in combination with its movement direction. During the skier's movement, it is continuously judged whether its trajectory enters the above-mentioned advertising reference plane. When the trajectory projection first enters the reference plane, the gaze duration recording mechanism is immediately activated. The system also begins tracking the changes in the skier's relative position to the ad space, as well as the angle between their gaze and the ad's normal. This angle is estimated by analyzing the angle between the skier's facial orientation and the ad surface, typically achieved through visual pose estimation models or head orientation detection technology. If the skier's viewing angle remains within the set focus range (i.e., the angle with the ad surface's normal is less than a threshold), and the skier's gaze duration in this state exceeds another threshold, the behavior data is marked as a valid gaze behavior, and a valid gaze data record is generated. The threshold setting here can be flexibly configured based on the actual advertising effectiveness requirements of the ski resort, balancing recognition accuracy and applicability.

[0033] Furthermore, the ad attention feature set is generated based on the aggregated effective gaze data, where the attention weight is added as the accompanying weight of the effective gaze data during the aggregation process.

[0034] As a preference of the above embodiment, firstly, effective gaze data of the skier is obtained from the above steps. The effective gaze data includes information such as the gaze duration, gaze angle, and distance between the skier and the ad space of the skier. After the initial acquisition of these data, they are cleaned and preprocessed to remove noise and outliers to ensure the accuracy of the data. In this embodiment, the attention weight is mainly adjusted based on the real-time distance between the skier and the ad space. Specifically, when the skier is closer to the ad space, the gaze attention weight is higher, and vice versa. The weight adjustment can take into account other factors, such as the skier's movement state and the potential impact of environmental factors on attention. When generating the advertising attention feature set, a preliminary weight is first assigned to each effective gaze data. At this time, the weight is mainly set based on the gaze duration and gaze angle. For example, If the skier's gaze duration is longer and the angle is more perpendicular to the ad space, the weight is higher. Subsequently, the weight is dynamically adjusted according to the distance between the skier and the ad space. For example, when the skier is farther away from the ad space, the weight of the gaze data is reduced, and vice versa. The aggregation process combines the attention weight of each valid gaze data with other features, and performs weighted averaging or other appropriate aggregation methods on this basis. In this way, the generated advertising attention feature set can accurately reflect the skier's attention to the ad space and provide a basis for subsequent advertising effect evaluation. Finally, the weighted aggregated advertising attention feature set will be used as output and enter the subsequent advertising effect analysis and optimization process. This feature set can help advertisers accurately evaluate the effectiveness of advertising, thereby making reasonable adjustments to advertising content, advertising location and delivery strategy.

[0035] Furthermore, we generate quantitative indicators of advertising effectiveness, including: After the effective gaze data in the advertising attention feature set is converted into the frequency domain, it is combined with the attention weight and historical behavior data to predict the probability of behavioral conversion guided by the advertisement, and output the quantitative indicators of the advertising effect.

[0036] As a preferred embodiment of the above, first, the advertising attention feature set generated in the above steps is obtained, including effective gaze data and corresponding attention weights. These effective gaze data represent the degree of attention of the skier to the advertisement, and the attention weight reflects the intensity of the skier's interest in the advertisement. In this step, frequency domain conversion (such as Fourier transform) is used to convert these time domain data into frequency domain data. The purpose of frequency domain conversion is to analyze the attention pattern of skiers to advertisements from different frequency levels, especially to identify which frequency gaze behaviors are more closely related to the advertising effect through periodic patterns; the gaze duration and gaze angle data are decomposed, and their frequency domain features are extracted respectively. The low-frequency components representing continuous attention and the high-frequency components representing intermittent or short-term attention in the frequency domain are obtained through Fourier transform. Frequency components, this processing helps to identify different types of attention behaviors; after the frequency domain conversion, the skier's attention weight and historical behavior data are combined to predict the probability of behavioral conversion guided by advertising. Historical behavior data includes the skier's past advertising viewing records, skiing routes, interactive behaviors and other information. Based on this data, machine learning algorithms can be used to predict the advertising effect; through the combination of the above-mentioned frequency domain conversion and behavior conversion probability prediction, a quantitative indicator of advertising effect is finally generated. This indicator can reflect the actual influence of the advertisement, including but not limited to the attractiveness of the advertisement, the ability of the advertisement to guide the skier's behavior, the performance of the advertisement in different environments or conditions, etc. This indicator can also be combined with other monitoring data of the ski resort to further improve the accuracy of the evaluation of advertising effect.

[0037] Furthermore, it also includes collecting real-time weather warning data. When the weather warning data reaches a threshold, a confidence correction coefficient is added to the quantitative indicators of advertising effectiveness. Example 2

[0038] Based on the same inventive concept as the ski resort outdoor advertising monitoring data analysis method in the aforementioned embodiment, the present invention further provides a ski resort outdoor advertising monitoring data analysis system, the system comprising: The data acquisition module collects multi-source monitoring data of outdoor advertising spaces at ski resorts, including video streams of advertising spaces, environmental sensor data, and skier behavior trajectories. The delayed image module performs dynamic environmental interference correction on the video stream of the advertising slot based on environmental sensor data to generate an advertising exposure image sequence; The feature generation module constructs the dynamic spatiotemporal association between skiers and ad spots based on the ad exposure image sequence and skier behavior trajectory, and extracts the ad attention feature set; The effect quantification module inputs the advertising attention feature set into the advertising effect evaluation model to generate advertising effect quantitative indicators.

[0039] The above-mentioned adjustment system in the present invention can effectively realize a method for analyzing outdoor advertising monitoring data in a ski resort. The technical effects that can be achieved are as described in the above-mentioned embodiments and will not be repeated here.

[0040] Furthermore, the feature generation module includes: A reference setting unit, which takes the visible area of ​​the advertisement exposure image sequence as a reference plane and projects the skier's behavior trajectory onto the reference plane; The gaze timing unit starts the gaze duration timing when the skier's trajectory enters the reference surface; The data recording unit records valid gaze data when the angle between the skier's behavior trajectory and the normal direction of the reference surface and the gaze duration are less than a threshold.

[0041] Similarly, the above-mentioned optimization schemes for the system can also respectively achieve the corresponding optimization effects of the method in Example 1, which will not be repeated here.

[0042] Although the present application has been described with reference to specific features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and drawings are merely illustrative of the present application as defined herein and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, the present application is intended to include such modifications and variations as fall within the scope of the present application and its equivalents.

Claims

1. A method for analyzing outdoor advertising monitoring data at a ski resort, characterized in that: The method comprises: Collecting multi-source monitoring data of outdoor advertising spaces at ski resorts, including video streams of advertising spaces, environmental sensor data, and skier behavior trajectories; Based on the environmental sensing data, dynamic environmental interference correction is performed on the video stream of the advertising position to generate an advertising exposure image sequence; Based on the advertisement exposure image sequence and the skier's behavior trajectory, a dynamic spatiotemporal association between the skier and the advertisement position is constructed, and an advertisement attention feature set is extracted; The advertisement attention feature set is input into an advertisement effect evaluation model to generate an advertisement effect quantitative index.

2. The ski resort outdoor advertising monitoring data analysis method according to claim 1, characterized in that: Generate an ad exposure image sequence, including: Acquiring the environmental sensor data, including light intensity, snow and fog concentration, and wind speed data; Constructing a brightness compensation matrix according to the light intensity to correct the advertising position video stream; Compensating for motion blur of the video stream of the advertising spot based on the wind speed data; A defogging convolution kernel is generated based on the snow fog concentration, and the advertising position video stream is corrected according to the defogging convolution kernel to generate the advertising exposure image sequence.

3. The ski resort outdoor advertising monitoring data analysis method according to claim 1, characterized in that: The generation of the skier's behavior trajectory includes: Get the skier's starting point on the trail and generate the initial position coordinates; Based on the real-time changes of the initial position coordinates, the skier's clothing color and ski equipment outline features corresponding to the moving area are obtained and combined to generate dynamic visual features; The skier's behavior trajectory is generated by combining the initial position coordinates with the dynamic visual features.

4. The ski resort outdoor advertising monitoring data analysis method according to claim 3, characterized in that: Merge to generate dynamic visual features, including: Performing spatial transformation on the color features of the skier's clothing to extract hue and saturation components to generate a first feature vector; Performing key point detection on the ski equipment contour features, obtaining a contour curvature histogram and generating a second eigenvector; combining the first feature vector and the second feature vector of each skier to output the dynamic visual feature; An attention weight is added to the combination of the first feature vector and the second feature vector, and the attention weight is dynamically adjusted according to the real-time distance between the skier and the advertising space.

5. The ski resort outdoor advertising monitoring data analysis method according to claim 1, characterized in that: The dynamic spatiotemporal association includes: Taking the visible area of ​​the advertisement exposure image sequence as a reference plane, projecting the skier's behavior trajectory onto the reference plane; When the skier's behavior trajectory enters the reference surface, the gaze duration timer is started; When the angle between the skier's behavior trajectory and the normal direction of the reference surface and the gaze duration are less than a threshold, it is recorded as valid gaze data.

6. The ski resort outdoor advertising monitoring data analysis method according to claim 5, characterized in that: The advertisement attention feature set is generated by aggregating the effective gaze data, wherein an attention weight is added as an accompanying weight of the effective gaze data during the aggregation process.

7. The ski resort outdoor advertising monitoring data analysis method according to claim 1, characterized in that: Generate quantitative indicators of advertising effectiveness, including: After the effective gaze data in the advertisement attention feature set is converted into the frequency domain, the probability of behavioral conversion guided by the advertisement is predicted by combining the attention weight and historical behavior data, and the quantitative index of the advertisement effect is output.

8. The ski resort outdoor advertising monitoring data analysis method according to claim 1, characterized in that: It also includes collecting real-time weather warning data, and when the weather warning data reaches a threshold, adding a confidence correction coefficient to the advertising effect quantitative index.

9. Ski resort outdoor advertising monitoring data analysis system, characterized by: The system comprises: The data acquisition module collects multi-source monitoring data of outdoor advertising spaces at ski resorts, including video streams of advertising spaces, environmental sensor data, and skier behavior trajectories. The delayed image module performs dynamic environmental interference correction on the video stream of the advertising slot based on environmental sensor data to generate an advertising exposure image sequence; The feature generation module constructs the dynamic spatiotemporal association between skiers and ad spots based on the ad exposure image sequence and skier behavior trajectory, and extracts the ad attention feature set; The effect quantification module inputs the advertising attention feature set into the advertising effect evaluation model to generate advertising effect quantitative indicators.

10. The ski resort outdoor advertising monitoring data analysis system according to claim 9, characterized in that: The feature generation module includes: A reference setting unit, which takes the visible area of ​​the advertisement exposure image sequence as a reference plane and projects the skier's behavior trajectory onto the reference plane; The gaze timing unit starts the gaze duration timing when the skier's trajectory enters the reference surface; The data recording unit records valid gaze data when the angle between the skier's behavior trajectory and the normal direction of the reference surface and the gaze duration are less than a threshold.

Images