Advertisement material effect evaluation method, device and equipment based on weighted fusion analysis

Abstract

The present application relates to the technical field of image processing, solve the problem that the prior art cannot associate the advertisement with the original material, the evaluation material in different scenes is low in accuracy, provide a kind of advertisement material effect evaluation method, device and equipment based on weighted fusion analysis.The method comprises: original video material is detected and blind watermark information is embedded, and target video highlight is obtained;Blind watermark information extraction is carried out on target video highlight, and material segment information is determined;According to material segment information, the collected data is statistically analyzed, and the real-time evaluation parameter and the original video material usage frequency are determined, the real-time evaluation parameter includes real-time material play number;When original video material is used multiple times, according to real-time material play number, the weighted average calculation of each single material evaluation result is carried out, and the multiple material evaluation result is determined.The present application provides accurate material performance data, which helps advertisers to optimize material use.

Description

[0001] The present application is a divisional application of the invention patent application with the application number 202411487805.3, the title of which is "Advertisement video material delivery effect evaluation method, device, equipment and storage medium", and which was filed on October 24, 2024. TECHNICAL FIELD

[0002] The present application relates to the technical field of image processing, and in particular to an advertisement material effect evaluation method, device and equipment based on weighted fusion analysis. BACKGROUND

[0003] Under the background of the rapid development of current digital advertising, advertisers are increasingly relying on diversified video materials for personalized and accurate delivery to improve user conversion rate and brand exposure. However, with the increase in the number and frequency of use of materials, how to scientifically and timely evaluate the performance of each original video material in actual delivery has become a key problem in advertising delivery optimization. Especially in the scene of mixed delivery of multiple materials and multiple versions, it is difficult to identify the advantages and disadvantages of specific materials by relying only on overall delivery effect, which may lead to resource waste and deviation of optimization direction.

[0004] However, the existing technology has the following limitations: material management usually relies on manual operation, which may cause repeated use of materials, confusion and loss of versions, and weak correlation between advertisement clips and original materials, resulting in inability to accurately track the use effect of materials; the monitoring and analysis of advertising effect mainly rely on overall data, lacking detailed analysis of the performance of specific materials.

[0005] The existing Chinese patent CN104954819A discloses a method for realizing intelligent monitoring and broadcasting of new media advertisements by using digital watermark encoding technology. The method includes: first, classifying and managing new media advertisements to be broadcast by an advertisement material management module, then inserting watermarks into the new media advertisements to be broadcast by a digital watermark encoding module, and performing unique coding identification, the terminal device transmits the results of user terminal analysis (containing unique codes of digital watermarks) to the national advertisement material library through the Internet or the bidirectional backhaul network of the broadcast network operator, then, comparing the collected user terminal advertisement analysis data with the front-end advertisement material library to obtain which new media advertisement content the user has watched, and the touch frequency of specific advertisements and other advertisement monitoring data. Although the above-mentioned patent discloses a technical solution for obtaining advertisement monitoring data by comparing advertisement analysis data with the material library, the correlation between advertisement clips and original materials is not strong, and once the materials are used in multiple clips or delivered on different platforms, it is difficult to track their specific performance and contribution, which limits the improvement and optimization of material utilization rate; and lacking detailed analysis of the performance of specific materials, which leads to the inability to accurately evaluate the effect of materials in different scenarios, thereby making it difficult to implement targeted optimization strategies.

[0006] To this end, how to associate the advertisement film with the original material and accurately evaluate the effect of the material in different scenes is a problem to be solved. SUMMARY

[0007] Therefore, the application provides an advertisement material effect evaluation method and device based on weighted fusion analysis to solve the problem that the advertisement film cannot be associated with the original material and the accuracy of evaluating the effect of the material in different scenes is low in the prior art.

[0008] The technical solution adopted by the application is as follows:

[0009] In a first aspect, the application provides an advertisement material effect evaluation method based on weighted fusion analysis, which comprises:

[0010] Scene switching detection and blind watermark information embedding are performed on at least one original video material in an advertisement launching scene to obtain a target video highlight;

[0011] Blind watermark information extraction is performed on the target video highlight, and according to the extracted blind watermark information, material segment information is determined, wherein the material segment information at least includes the original video material name and the original video material position used in the target video highlight;

[0012] According to the material segment information, second-level statistics are performed on the collected launching data to determine real-time evaluation parameters related to the original video material evaluation and the number of uses of the original video material in the advertisement launching, wherein the real-time evaluation parameters at least include real-time material play number, real-time material click rate and real-time material churn rate;

[0013] According to the number of uses, when the original video material is used multiple times, according to the real-time material play number, weighted average calculation is performed on the single material evaluation result corresponding to each original video material to determine the multiple material evaluation result.

[0014] Preferably, the scene switching detection and blind watermark information embedding are performed on at least one original video material in an advertisement launching scene to obtain a target video highlight, which comprises:

[0015] The original video material is decomposed into multiple frames of original video images, and the color histogram corresponding to each original video image is obtained;

[0016] Difference value calculation is performed on the color histograms of adjacent frames of original video images to obtain a color histogram difference value;

[0017] If the color histogram difference value is greater than a preset color histogram threshold value, the adjacent frames of original video images are classified into two scene video segments;

[0018] The preset blind watermark information is embedded into each frame image in each scene video segment by using the image processing technology based on the frequency domain transform, and each scene video segment after embedding the blind watermark information is edited to obtain the target video highlight.

[0019] Preferably, the embedding of the preset blind watermark information into each frame image in each scene video segment by using the image processing technology based on the frequency domain transform and the editing of each scene video segment after embedding the blind watermark information to obtain the target video highlight comprises:

[0020] The color space conversion is performed on each frame image in each scene video segment to obtain a target color channel.

[0021] The discrete cosine transform is performed on the target color channel to convert each frame image in each scene video segment into a scene frequency domain image.

[0022] The high-frequency region in the scene frequency domain image is modified according to a preset key, and the preset blind watermark information is embedded into the scene frequency domain image.

[0023] Preferably, the preset color histogram threshold is determined by the following steps:

[0024] A current frame image in adjacent frame original video images is obtained.

[0025] According to the current frame image, a color histogram set of all frame images before the current frame image is determined.

[0026] According to the color histogram set, a preset smoothing parameter and an initial threshold are combined to determine the color histogram threshold.

[0027] Preferably, the modification of the high-frequency region in the scene frequency domain image according to the preset key and the embedding of the preset blind watermark information into the scene frequency domain image comprise:

[0028] According to the preset blind watermark information, a template size of a region to be embedded is obtained.

[0029] According to the template size of the region to be embedded, a target region in the scene frequency domain image is selected, wherein the target region has the same size as the template size of the region to be embedded.

[0030] The frequency domain coefficient analysis is performed on the target region, and the sum of squares of all frequency domain coefficients in the target region is calculated as a region energy value.

[0031] The region energy value is compared with a preset energy threshold, and if the region energy value is greater than the energy threshold, the target region is classified as a high-frequency region.

[0032] if the region energy value is less than or equal to the energy threshold value, the target region is classified as a low-frequency region;

[0033] The frequency domain coefficient analysis of any target region in the scene frequency domain image is repeated until all regions in the scene frequency domain image are classified as high-frequency regions or low-frequency regions.

[0034] A plurality of scene frequency domain images are acquired, and a scene frequency domain image is selected as a current frame image;

[0035] The positions of each pixel point in the high-frequency region in the current frame image and the positions of each pixel point in the previous frame image are analyzed using an optical flow method to determine the optical flow field corresponding to each pixel point;

[0036] The optical flow field corresponding to each pixel point is subjected to mean value calculation to determine the optical flow mean value corresponding to the high-frequency region in the current frame image;

[0037] When the optical flow mean value is less than a preset optical flow threshold value, the high-frequency region in the current frame image is modified, and the preset blind watermark information is embedded in the scene frequency domain image;

[0038] The selection of a scene frequency domain image as a current frame image for blind watermark information embedding is repeated until the preset blind watermark information is embedded in all scene frequency domain images.

[0039] Preferably, after the use frequency, when the original video material is used multiple times, the single-material evaluation result corresponding to each time of original video material is weighted and averaged according to the real-time material play number to determine the multiple-material evaluation result, and the method further comprises:

[0040] An advertisement type and a delivery platform when an advertisement is delivered based on the target video highlight are acquired;

[0041] According to the advertisement type and the delivery platform, the material evaluation result is corrected again, and the corrected evaluation result is used as the delivery effect analysis result.

[0042] Preferably, the single-material evaluation result is determined by the following steps:

[0043] The total duration of the target video highlight is acquired, and the material position gear is determined according to the original video material position and the total duration, the material position gear being used to identify the relative time interval of the original video material in the target video highlight;

[0044] According to the material position gear, a standard parameter is determined, wherein the standard parameter at least includes a standard material play number, a standard material click rate, and a standard material churn rate;

[0045] The standard parameter and the evaluation parameter are calculated to determine the single material evaluation result.

[0046] In a second aspect, the application provides an advertisement material effect evaluation device based on weighted fusion analysis, which comprises:

[0047] A target video highlight determination module is configured to perform scene switching detection and blind watermark information embedding on at least one original video material in an advertisement delivery scenario to obtain a target video highlight.

[0048] A blind watermark information extraction module is configured to extract blind watermark information from the target video highlight, and determine material segment information according to the extracted blind watermark information, wherein the material segment information at least includes a name and a position of the original video material used in the target video highlight.

[0049] A delivery data statistics module is configured to perform second-level statistics on pre-collected delivery data according to the material segment information, and determine real-time evaluation parameters related to original video material evaluation and a usage frequency of the original video material in advertisement delivery, wherein the real-time evaluation parameters at least include real-time material play number, real-time material click rate and real-time material churn rate.

[0050] A multiple use material evaluation module is configured to, when the original video material is used multiple times, perform weighted average calculation on single material evaluation results corresponding to each time of the original video material according to the real-time material play number, and determine a multiple material evaluation result.

[0051] In a third aspect, the application further provides an electronic device, which comprises at least one processor, at least one memory and computer program instructions stored in the memory, and when the computer program instructions are executed by the processor, the method of the first aspect in the above-mentioned embodiments is implemented.

[0052] In a fourth aspect, the application further provides a storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the method of the first aspect in the above-mentioned embodiments is implemented.

[0053] In summary, the application has the following beneficial effects:

[0054] The application provides an advertisement material effect evaluation method, device and equipment based on weighted fusion analysis, which comprises the following steps: acquiring at least one original video material uploaded by a target object in an advertisement launching scene; performing scene switching detection on the original video material, and cutting the original video material into multiple scene video clips; embedding preset blind watermark information into each frame image in each scene video clip by using an image processing technology based on frequency domain transformation; performing editing on each scene video clip after the blind watermark information is embedded to obtain a target video highlight; extracting the blind watermark information from the target video highlight, and determining material clip information according to the extracted blind watermark information, wherein the material clip information at least comprises a name and a position of the original video material used in the target video highlight; when the target video highlight is launched, evaluating each material clip according to collected launching data information and the material clip information, and summarizing all material evaluation results to obtain a launching effect evaluation result. By introducing a blind watermark information embedding and extracting mechanism in the advertisement launching scene, the application solves the problem that it is difficult to accurately associate an advertisement with original video material in the prior art. When the target video highlight is generated, scene switching detection is performed on the original video material, and unique blind watermark information is embedded, so that each material still has traceability after being synthesized. Then, the original material name and the specific position in the highlight can be accurately identified by extracting the blind watermark information from the launched advertisement. This process realizes automatic positioning and identification of the material clip, so that the system can perform second-level statistical analysis on the playing, clicking and loss data of each material in different scenes at a specific launching time. By combining these real-time evaluation parameters with the use frequency of the material, the scheme further supports weighted average evaluation of the material used multiple times, effectively improves the evaluation precision of the material launching effect, provides more targeted and effective optimization basis for the advertiser, and significantly overcomes the limitations of the prior art in coarse material evaluation granularity and poor accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed to be used in the embodiments of the application will be briefly introduced. For those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings, and these are within the protection scope of the application.

[0056] Figure 1 The flowchart of the whole work of the advertisement material effect evaluation method based on weighted fusion analysis in the embodiment 1 of the application is shown in the figure.

[0057] Figure 2 The flowchart of the scene switching detection on the original video material in the embodiment 1 of the application is shown in the figure.

[0058] Figure 3 Flowchart for determining the color histogram threshold in Embodiment 1 of the present application;

[0059] Figure 4 Flowchart for embedding preset blind watermark information into each frame image in each scene video segment in Embodiment 1 of the present application;

[0060] Figure 5 Flowchart for extracting blind watermark information from the target video highlight set which is clipped from each scene video segment after embedding blind watermark information in Embodiment 1 of the present application;

[0061] Figure 6 Flowchart for evaluating each material segment in Embodiment 1 of the present application;

[0062] Figure 7 Flowchart for determining single material evaluation result in Embodiment 1 of the present application;

[0063] Figure 8 Structure block diagram of the advertisement material effect evaluation device based on weighted fusion analysis in Embodiment 2 of the present application;

[0064] Figure 9 Structure diagram of the electronic device in Embodiment 3 of the present application. DETAILED DESCRIPTION

[0065] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. It should be noted that, in this document, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. In the description of the present application, it should be understood that the orientations or positional relationships indicated by the terms “center”, “upper”, “lower”, “front”, “back”, “left”, “right”, “vertical”, “horizontal”, “top”, “bottom”, “inner”, “outer” and the like are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. Moreover, the terms “include”, “contain” or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or device. Without more limitations, the elements defined by the statement “include” do not exclude the presence of additional identical elements in the process, method, article or device that includes the elements. If there is no conflict, the embodiments of the present application and the various features in the embodiments can be combined with each other, and are all within the protection scope of the present application.

[0066] Embodiment 1

[0067] See Figure 1 Embodiment 1 of the present application discloses an advertisement material effect evaluation method based on weighted fusion analysis, which comprises:

[0068] S1: obtaining at least one original video material uploaded by a target object under an advertisement launching scene;

[0069] Specifically, obtaining the original video material uploaded by the target object under the advertisement launching scene means that a video file uploaded by a user is first received and used as initial input data. The original video material contains various types of content, such as product advertisements, brand promotions, film and television clips, and user-generated content (UGC). The original video material covers different scenes, themes and styles, and has diversity and complexity. By further processing these original video materials, a foundation is laid for subsequent material segment identification and evaluation, so that when the original video material is edited and launched, the use and launching effect of each material segment can be accurately extracted and analyzed.

[0070] S2: scene cut detection is performed on the original video material, and the original video material is cut into a plurality of scene video clips according to the scene cut detection result;

[0071] Specifically, the scene cut detection performed on the original video material refers to automatically identifying the transition points between different scenes in the video, such as the moment when a scene is switched to another scene, by using image processing and computer vision technology, and the original video material is cut into a plurality of scene video clips according to the scene cut detection result, each clip representing a continuous visual and content unit. This process not only accurately locates the start and end positions of each scene, but also provides a clear segmentation basis for subsequent blind watermark embedding and material management, thereby ensuring that each clip can be independently and accurately tracked and analyzed in subsequent editing, delivery and effect evaluation. In this way, by refining the processing of video material, each scene clip can be more efficiently managed and utilized, and the accuracy of overall advertisement delivery and effect evaluation can be improved.

[0072] In an embodiment, referring to Figure 2 , the S2 includes:

[0073] S21: the original video material is decomposed into a plurality of original video images, and a color histogram corresponding to each original video image is obtained;

[0074] Specifically, decomposing the original video material into a plurality of original video images is the first step of scene cut detection. First, the continuous original video material is decomposed into individual original video images, and each second of video usually contains a plurality of frames (such as 30 frames per second). Each frame of image represents an instantaneous picture in the video. Next, a color histogram is calculated for each frame of original video image. The color histogram is a statistical chart representing the number of pixels of each color in the image. By calculating the color histogram (denoted as H_i) of each frame, the color distribution characteristics of each frame of original video image can be obtained, which are used in subsequent scene cut detection.

[0075] S22: difference calculation is performed on the color histograms of adjacent frames of original video images to obtain a color histogram difference value;

[0076] Specifically, after obtaining the color histogram of each original video image, the color histogram difference between adjacent frames is calculated, which is achieved by subtracting the color histogram of adjacent frames. Common methods include relative entropy (KL divergence) or L2 norm. For frames F_i and F_{i+1}, the color histogram difference D_i = d(H_i, H_{i+1}) is calculated, where d can be KL divergence or L2 norm. In this way, the degree of change in color distribution between adjacent frames is quantified. The calculation of the color histogram difference is an important basis for detecting scene changes, because scene changes are usually accompanied by significant changes in image color.

[0077] S23: If the color histogram difference value is greater than the preset color histogram threshold, the adjacent frame original video image is classified into two scene video segments.

[0078] Specifically, to determine whether there is a scene change, the calculated color histogram difference value D_i of each frame is compared with the preset color histogram threshold T_i. To improve the robustness of detection, an adaptive threshold is used, which is dynamically adjusted according to the current frame. If the color histogram difference value D_i of a certain frame is greater than the adaptive threshold T_i, it is considered that there is a scene change between frames F_i and F_{i+1}. At this time, the adjacent frame original video image is divided into two different scene video segments, thereby realizing scene cutting of the video material. This method ensures that each scene video segment has consistent visual features and is significantly different from other scenes.

[0079] In an embodiment, referring to Figure 3 , the preset color histogram threshold is determined by the following steps:

[0080] S2301: Obtain a current frame image in adjacent frame original video images;

[0081] Specifically, in the process of scene change detection, the original video material is processed frame by frame. First, the current frame image in adjacent frame original video images, i.e., the frame being processed, is obtained. This frame image is the target of the current analysis. By calculating the color histogram difference between this frame and the previous frame, it can be determined whether there is a scene change.

[0082] S2302: Determine a color histogram set of all frame images before the current frame image according to the current frame image;

[0083] Specifically, after obtaining the current frame image, a color histogram set of all frame images before the current frame image is collected. This set contains the color histograms of each frame from the beginning of the video to the current frame (H_1, H_2,..., H_{i-1}). These histograms represent the color distribution of each frame image, and by analyzing these data, the color change in the first few frames of the video can be understood. This information is very important for calculating the change trend of the current frame and subsequent threshold adjustment.

[0084] S2303: According to the color histogram set, a preset smoothing parameter and an initial threshold value are combined to determine the color histogram threshold value.

[0085] Specifically, in order to make the detection more robust, an adaptive threshold is used to determine whether there is a scene cut. According to the color histogram set, a preset smoothing parameter (a) and an initial threshold value (T), the threshold value (T_i) of the current frame is dynamically adjusted. The specific calculation method is T_i = aT + (1-a)M_i, where M_i is the average difference of the previous n frames, represented as M_i = (1 / n)Σ_{j=i-n+1}^{i}D_j. In this way, the threshold value can be dynamically adjusted according to the change of the current frame, making the scene cut detection more sensitive and accurate. If the color histogram difference D_i between the current frame and the previous frame is greater than the adaptive threshold T_i, the system determines that there is a scene cut between the two frames. The setting of this adaptive threshold ensures that it can be flexibly adjusted when processing different video materials, improving the robustness and accuracy of the detection.

[0086] In an embodiment, the S23 further comprises:

[0087] S231: If the color histogram difference value is greater than the preset color histogram threshold value, the corresponding adjacent audio frames of the adjacent frame original video images are obtained.

[0088] Specifically, when the color histogram difference value of adjacent video frames exceeds the preset color histogram threshold value, it is preliminarily judged that there may be a scene cut. At this time, in addition to relying on image information, audio information needs to be further verified, therefore, the audio frames corresponding to the two video frames are synchronously extracted, the audio frame refers to a piece of audio data aligned with the video frame on the time axis, by obtaining these audio frames, the change of the audio signal is analyzed to confirm whether a scene transition really occurs, this process ensures the synchronous processing of video and audio information, laying a foundation for subsequent multi-modal analysis.

[0089] S232: Using a preset audio feature extraction algorithm, audio analysis is performed on the adjacent audio frames to determine the audio feature change amplitude between the adjacent audio frames.

[0090] Specifically, after obtaining the audio frames, the adjacent audio frames are analyzed to detect changes in the audio using audio feature extraction algorithms, including spectral analysis, pitch variation detection, audio energy analysis, frequency band energy variation, and audio event detection techniques. By using the preset audio feature extraction algorithm, significant changes in the audio are identified, such as the switching of background music, the start or end of a conversation, or sudden environmental sound changes. By comparing the features of adjacent audio frames, the audio feature change amplitude is calculated, which can reflect the intensity of the change in the audio content within a short period of time, thereby providing auxiliary basis for judging scene switching. If the audio feature changes significantly, it means that in addition to the visual scene, the audio environment has also changed significantly, which further supports the judgment of scene switching.

[0091] S233: inputting the adjacent frame original video image and the adjacent audio frame into a pre-trained multi-modal large model to output scene-related text content information;

[0092] Specifically, the adjacent frame original video image is input into a pre-trained multi-modal large model for processing. The multi-modal large model can comprehensively process visual and audio information and automatically generate scene-related text descriptions or extract text information in the scene through deep learning technology. This step includes optical character recognition (OCR) to detect and extract text content in the video frame, such as identification, subtitles, or written information in the scene. In addition, the multi-modal large model can also analyze the combination of visual and audio features of the scene to generate descriptive text information. These text information may include explanations of scene content, capture of semantic changes, or even transcription of detected conversations or important sounds. By generating or extracting these text information, the scene content can be further understood at the semantic level, providing more rich information for scene switching.

[0093] S234: weighting analysis of the audio feature change amplitude and the text content according to a preset weight, and determining whether to divide the adjacent frame original video image into two scene video segments according to the weighting analysis result.

[0094] Specifically, after the extraction of audio and text information is completed, the audio feature change amplitude and the text content are analyzed by weighting according to a preset weight, and the purpose of the weighted analysis is to comprehensively analyze the multi-modal information and make a more accurate scene switching judgment. The preset weight can be adjusted according to the characteristics of the video content and the analysis target. For example, for a music video, the weight of the audio feature can be higher, and for a video with rich dialogue scenes, the weight of the text content can be higher. Through this weighted analysis, the audio, text and image information are combined, and the change amplitudes between them are comprehensively considered. If the weighted analysis result shows a significant change, it is judged that there is a scene switching, and the adjacent video frames are divided into two different scene video segments. In this way, not only the visual features are used, but also the audio and text information are fully utilized, so that the detection of scene switching is more accurate and reliable.

[0095] S3: using image processing technology based on frequency domain transformation, embedding preset blind watermark information into each frame image in each scene video segment, and editing each scene video segment after embedding the blind watermark information to obtain a target video highlight;

[0096] In an embodiment, referring to Figure 4 , the S3 comprises:

[0097] S31: performing color space conversion on each frame image in each scene video segment to obtain a target color channel;

[0098] Specifically, each frame image in each scene video segment is converted from RGB color space to YUV color space. YUV color space is divided into three channels: Y channel (luminance), U channel (chrominance), and V channel (density). The converted YUV image can more effectively separate color information and luminance information, so that we can operate on the U and V channels without affecting the overall visual quality. Through this conversion, the target color channel (U and V) of each frame image is obtained, laying a foundation for subsequent blind watermark embedding.

[0099] S32: performing discrete cosine transform on the target color channel to convert each frame image in each scene video segment into a scene frequency domain image;

[0100] Specifically, after obtaining the target color channel, discrete cosine transform (DCT) is performed on these channels. DCT is a frequency domain transformation method that can convert an image from spatial domain to frequency domain. Through DCT, the image is represented as a set of frequency components, where the low-frequency components contain the main image information, and the high-frequency components contain less detailed information. This conversion allows us to process in the frequency domain to hide the blind watermark information while minimizing the impact on the visual quality of the image. After DCT processing, the U and V channels of each frame image are converted into frequency domain images, showing the weights of each frequency component.

[0101] S33: modifying a high-frequency region in the frequency domain image of the scene according to a preset key, and embedding preset blind watermark information in the frequency domain image of the scene.

[0102] Specifically, a preset key is used to determine the specific position of embedding the blind watermark information in the frequency domain image. The key is a set of parameters for positioning, which ensures that the position of the blind watermark embedding is consistent and traceable for each frame. The high-frequency region in the frequency domain image of the scene is selected for modification processing because these regions contain less visual information and have the least impact on the overall visual effect of the image after modification. By embedding blind watermark information (i.e., a unique ID) in the high-frequency region, it is ensured that each frame of image contains identifiable covert information. This process converts the frequency domain image back to the spatial domain image through inverse transformation (IDCT), and finally completes the embedding of blind watermark information. In this way, each frame of image has invisible identification information, ensuring that the material can be accurately identified and traced when it is cited in a film.

[0103] In an embodiment, the S33 comprises:

[0104] S331: obtaining a region template size to be embedded according to preset blind watermark information;

[0105] Specifically, the information content of the blind watermark, such as text, image or other forms of data, needs to be determined first. According to the complexity and information amount of the watermark, a suitable region template size is calculated. For example, if the watermark is a short text, only an 8x8 or 16x16 template size is needed, while if the watermark is an image or larger data, a larger region is needed. The selection of the template size affects the clarity and visibility of the watermark. Too large a template size will cause a significant impact on the image after embedding, and too small a template size will not be able to completely transmit the information. Therefore, a reasonable template size can improve the concealment and transmission efficiency of the watermark.

[0106] S332: selecting a target region in the frequency domain image of the scene according to the region template size to be embedded, wherein the target region has the same size as the region template size to be embedded;

[0107] Specifically, a target region with the same size as the template size is selected from the frequency domain image of the scene. The selection method includes random selection or selection based on image features. For example, if a frequency domain image of a static scene is selected, an 8x8 region is randomly selected from it. In actual application, the selection of the target region takes into account the content characteristics of the image, such as selecting a region with a relatively flat background to ensure the concealment of the watermark. If the position of the target region happens to be in a part with less texture variation, the visual effect after embedding the watermark will be better.

[0108] S333: performing frequency domain coefficient analysis on the target region, and calculating a square sum of all frequency domain coefficients in the target region as a region energy value;

[0109] Specifically, frequency domain coefficient analysis is performed on the selected target region, and an energy value reflecting the intensity of the region is obtained by calculating the square sum of all frequency domain coefficients in the region. For example, when an 8x8 DCT coefficient matrix is analyzed, the square sum of each frequency domain coefficient in the matrix is calculated to obtain the energy value of the region. This value can indicate the details and intensity of the region, and is usually used to evaluate whether the region is suitable for embedding a watermark. For example, if the energy value of a region is 50, it means that the region is relatively weak.

[0110] S334: comparing the region energy value with a preset energy threshold value, and if the region energy value is greater than the energy threshold value, classifying the target region as a high-frequency region;

[0111] Specifically, the calculated region energy value is compared with a preset energy threshold value. If the region energy value is greater than the threshold value, the region is marked as a high-frequency region. For example, assuming that the preset energy threshold value is 40, if the energy value of a target region is 45, it means that the region has strong details and is suitable for carrying a watermark. Conversely, if the energy value is lower than the threshold value, it is not suitable for watermark embedding. Through such classification, the system can effectively identify which regions can effectively carry watermark information without compromising the overall quality of the image.

[0112] S335: if the region energy value is less than or equal to the energy threshold value, classifying the target region as a low-frequency region;

[0113] Specifically, when the region energy value is less than or equal to the preset energy threshold value, the region is marked as a low-frequency region. Low-frequency regions usually contain fewer details and are more flat, making them suitable for embedding watermarks without causing significant visual interference. For example, if the energy value is 35, it means that the region is relatively stable and suitable for carrying a watermark. Through such classification, a wise choice can be made between high-frequency regions and low-frequency regions to ensure the visual concealment of the watermark.

[0114] S336: repeating steps S332 to S335 until all regions in the scene frequency domain image are classified as high-frequency regions or low-frequency regions;

[0115] Specifically, a comprehensive analysis is performed on the entire scene frequency domain image, and all target regions are evaluated one by one through repeated steps S332 to S335, and the purpose of this is to ensure that each region is classified so that the subsequent watermark embedding strategy can be based on comprehensive data, for example, assuming that there are multiple 8x8 regions in an image, the regions are analyzed one by one until all regions are marked as high frequency or low frequency. This process ensures that the image can be comprehensively evaluated to select the most appropriate regions for watermark embedding.

[0116] S337: Obtain multiple scene frequency domain images, and select a scene frequency domain image as a current frame image;

[0117] Specifically, the current frame image is obtained from multiple scene frequency domain images, and the multiple frames provide dynamic background information for watermark embedding, which helps to improve the robustness of the watermark. For example, in a video sequence, there can be dozens of frames, and one frame is selected as the current processing object from these frames, which can better analyze the motion of the current frame and compare it with the previous frame to facilitate the embedding of dynamic watermarks.

[0118] S338: Analyze the positions of each pixel point in the high-frequency region in the current frame image and the positions of each pixel point in the previous frame image using the optical flow method to determine the optical flow field corresponding to each pixel point;

[0119] Specifically, the optical flow method is used to analyze the motion of the high-frequency region in the current frame. The optical flow method can effectively extract motion information by calculating the change in pixel position between adjacent frames. For example, if a pixel point in the high-frequency region of the current frame moves 5 pixels to the left, and the pixel point at the corresponding position in the previous frame remains unchanged, this motion information will be recorded and an optical flow field will be constructed. This process provides a basis for subsequent watermark embedding, identifying which regions have less motion and are therefore more suitable for embedding watermarks.

[0120] S339: Perform mean value calculation on the optical flow field corresponding to each pixel point to determine the optical flow mean value corresponding to the high-frequency region in the current frame image;

[0121] Specifically, the mean value of all high-frequency region pixel points in the optical flow field is calculated to determine the overall motion state of the region. For example, if the optical flow values of multiple pixel points in the high-frequency region are 2, 3, 4, and 5, the mean value is 3.5. This mean value can reflect the degree of motion of the region, and if the mean value is small, it indicates that the motion of the region is relatively small and is suitable for watermark embedding.

[0122] S3310: When the optical flow mean value is less than a preset optical flow threshold value, modify the high-frequency region in the current frame image and embed the preset blind watermark information in the scene frequency domain image;

[0123] Specifically, if the calculated average optical flow is lower than the preset optical flow threshold, the high-frequency region of the current frame will be modified to embed blind watermark information. For example, if the set optical flow threshold is 4 and the average is 3.5, it will be judged that the region is suitable for watermark embedding. At this time, the watermark information will be added to the frequency domain coefficients of the region, ensuring the concealment and effectiveness of the watermark.

[0124] S3311: Repeat steps S337 to S3310 until the preset blind watermark information is embedded in all scene frequency domain images.

[0125] Specifically, finally, steps S337 to S3310 are executed in a loop to process each scene frequency domain image frame by frame until all images complete watermark embedding. This continuous processing method can ensure that the watermark information is effectively transmitted in multiple frames of images and improve the robustness and concealment of the watermark. By dynamic analysis in each frame, it ensures that the watermark information can maintain consistency and stability in the entire video sequence.

[0126] S4: Blind watermark information extraction is performed on the target video highlight, and according to the extracted blind watermark information, material segment information is determined, wherein the material segment information at least includes the name and position of the original video material used in the target video highlight;

[0127] Specifically, first, each frame of image in the target video highlight is processed, the image is converted from the frequency domain to the spatial domain through inverse discrete cosine transform (IDCT), and the blind watermark information embedded in the high-frequency region is extracted. Using a preset key, the specific position and content of the blind watermark information in the image are accurately located. Through the extracted blind watermark information, the system can identify the unique ID in each frame of image, and determine which original material segments are used in the target video highlight. This process enables accurate tracking of the source and position of each material segment even if the material is edited, mixed, or repeatedly used in different scenes, ensuring that the reference of the material in the film can be perfectly identified, and achieving precise data backflow.

[0128] In an embodiment, referring to Figure 5 , the S4 includes:

[0129] S41: The target video highlight is decomposed into multiple target video images, and discrete cosine transform is performed on each target video image to determine a target frequency domain image.

[0130] Specifically, the target video highlight is decomposed into individual target video images, each frame representing a moment in the video, and a discrete cosine transform (DCT) is applied to each frame to convert the target video images from the spatial domain to the frequency domain, determining target frequency domain images. Through DCT, the image is decomposed into a series of frequency components, each representing different frequency changes in the image. These target frequency domain images reveal the high and low frequency information of the image, providing a basis for subsequent blind watermark extraction. Through this conversion, the system can process and analyze hidden information in the image in the frequency domain.

[0131] S42: Determine the target position of the blind watermark information in the target frequency domain image according to the key.

[0132] Specifically, after obtaining the target frequency domain images, a preset key is used to determine the specific position of the blind watermark information in these frequency domain images. The key is a set of predefined parameters used to identify the frequency components and positions where the blind watermark information is embedded. By referring to the key, the area in the frequency domain image containing the blind watermark information can be accurately found. This step ensures the accuracy of blind watermark extraction, as only the key can correctly decode the embedded hidden information. The use of the key enables systematic positioning of the unique ID information hidden in each frame of the image.

[0133] S43: Extract and restore the blind watermark information from the target frequency domain image according to the target position, and determine the material segment information.

[0134] Specifically, after determining the target position of the blind watermark information, the positions are processed to extract the hidden blind watermark information. The extraction process includes reading the modified information in the high-frequency area from the frequency domain image and restoring the frequency domain information to the spatial domain through inverse discrete cosine transform (IDCT). The extracted blind watermark information contains embedded unique IDs, which correspond one-to-one with the original material segments. Through these IDs, it can be identified and confirmed which original material segments are used in the target video highlight, ensuring the accuracy of material tracking and data backflow. This way, detailed usage and performance data can be provided for each material segment, optimizing advertising placement and content management.

[0135] S5: When the target video highlight is placed for advertising, evaluate each material segment based on the pre-collected placement data information and the material segment information, and summarize all material evaluation results to obtain the placement effect evaluation results.

[0136] Specifically, when the target video highlights are advertised, the collected advertising data is associated and analyzed with the previously extracted material segment information. The advertising data includes key indicators such as the number of plays, click rates, viewing time, user interaction, etc. By matching these advertising data with the unique ID of each material segment, the performance of each material segment in different advertising scenarios can be accurately evaluated. During the evaluation process, all indicators are considered to quantitatively score the effect of the material segment, and a detailed evaluation report is generated. Ultimately, the material evaluation results are output to provide targeted optimization suggestions for advertisers, helping them improve the utilization efficiency of materials and advertising effects in future advertising creation and placement.

[0137] In an embodiment, referring to Figure 6 , the S5 comprises:

[0138] S51: According to the material segment information, the pre-collected advertising data is statistically analyzed by seconds to determine real-time evaluation parameters related to the original video material evaluation and the number of times the original video material is used in advertising, wherein the real-time evaluation parameters at least include real-time material play count, real-time material click rate and real-time material churn rate.

[0139] Specifically, according to the material segment information, which at least includes the name and location of the original video material used in the target video highlights, the collected advertising data is processed in detail and analyzed by seconds. Specifically, for each original video material segment, first determine its specific use period in the video, for example, assume that the original video material segment is used between the 2nd second and the 5th second of the video, then focus on the advertising data within these seconds. For real-time material play count, the play count at the 2nd second when the material starts to be used is used as the evaluation benchmark. Next, calculate the average churn rate of the original video material segment per second within its use period. Real-time material churn rate refers to the proportion of viewers who leave the video at a certain second. Calculate the churn rate from the 2nd second to the 5th second, then take the average of these four values to get the average churn rate per second of the material. Similarly, the click rate per second is also calculated. Real-time material click rate reflects the audience's interaction with the advertisement during the video viewing process. The click rate from the 2nd second to the 5th second is counted separately, and the average value is calculated to get the average click rate per second of the material. Through such second-level statistics, key evaluation parameters related to material evaluation can be extracted, including play count, average churn rate and average click rate. These parameters provide basic data for subsequent material evaluation, making the evaluation process more accurate and scientific, and helping advertisers understand the specific performance of each material in different time periods and advertising scenarios, and optimize advertising placement strategies.

[0140] S52: According to the number of uses, if the original video material is used once, a single material evaluation result is determined according to the evaluation parameter and the original video material position;

[0141] Specifically, according to the number of uses, when the original video material is used once, the single evaluation result thereof is calculated by using the previously collected evaluation parameter and original video material position. This step ensures detailed evaluation of the single use of each material for subsequent comprehensive evaluation.

[0142] In an embodiment, referring to Figure 7 , the S52 comprises:

[0143] S521: Obtain the total duration of the target video highlight, and determine the material position gear according to the original video material position and the total duration;

[0144] Specifically, according to the material segment information, the specific use position of the material in the target video highlight is determined, which includes the start time and end time of the material and the specific time period in which the material is located in the video. This step helps to understand the position of the material in the video, providing a reference for subsequent evaluation. According to the position of the material segment in the video and the total duration of the video, the material position is divided into different gears. For example, the video can be divided into five gears: beginning, front, middle, back and end. Assuming that the total duration of the video is 35 seconds, each gear is about 7 seconds. The material center position is assumed to be 12 seconds to 15.3 seconds, and the average is 13.65. The material center position is calculated and rounded down. According to the center position of the material segment, the specific gear to which it belongs is calculated and determined.

[0145] S522: Determine the standard parameter according to the material position gear, wherein the standard parameter at least includes standard material play number, standard material click rate and standard material churn rate;

[0146] Specifically, after determining the specific gear of the material, the standard parameter is determined according to the material position gear, wherein the standard parameter at least includes standard material play number, standard material click rate and standard material churn rate; and each gear corresponds to different standard parameters, for example, five different position reference indexes are defined respectively, R[0].ctr represents the standard click rate of the beginning material, R[4].play represents the standard play number of the end material, and the standard parameter provides a benchmark for subsequent evaluation, making the evaluation process more accurate and targeted.

[0147] S523: Calculate the standard parameter and the evaluation parameter to determine the single material evaluation result. Specifically, finally, the evaluation parameter of the material is compared and analyzed with the standard parameter, and the evaluation result of the single use of the material is calculated. The evaluation result includes the scores of various indicators, and the performances of the play quantity, the click rate and the loss rate are comprehensively considered to finally obtain a comprehensive score. The calculation formula is:

[0148]

[0149] Wherein ε=0.000001, x.score represents the comprehensive score, x.play represents the play quantity of the original video material, x.ctr represents the average click rate per second of the original video material, x.lr represents the average loss rate per second of the original video material, x.pos represents the position of the original video material, R[x.pos].play represents the standard play quantity of the original video material at the position, R[x.pos].ctr represents the standard click rate of the original video material at the position, R[x.pos].lr represents the standard loss rate of the original video material at the position, and the symbol represents the lower limit a and the upper limit b of the truncated value. This score reflects the overall effect of the material in single use, and provides specific evaluation basis for the advertiser.

[0150] S53: If the original video material is used multiple times, the single material evaluation result corresponding to each time of use of the original video material is weighted and averaged according to the real-time material play quantity to determine the multiple material evaluation result.

[0151] Specifically, in the case that the material is used multiple times, the evaluation results of each use of the material are weighted and averaged. The basis for weighting is the logarithmic play quantity of each use, so as to ensure that the use times with larger play quantity have higher weight in the comprehensive evaluation. The calculation formula is as follows:

[0152]

[0153] Wherein X[i].play represents the play quantity of the original video material corresponding to the i-th use of the original video material; X[i].score represents the single material evaluation result corresponding to the i-th use of the original video material; in this way, a more comprehensive evaluation result can be obtained to reflect the overall performance of the material in multiple uses. This weighted average calculation method improves the accuracy of the evaluation and helps the advertiser to better understand the comprehensive effect of the material in different delivery scenarios.

[0154] After the S5, it further includes:

[0155] S61: Obtain the advertisement type and the delivery platform when the target video highlight is delivered.

[0156] Specifically, the target video highlights are obtained for the advertising platform and the advertising type that the advertisement faces when the advertisement is put on, for example, for traditional media such as television, when the material evaluation is carried out, the factors such as channel switching time and audience viewing habits are considered, and for streaming media platforms such as short video platforms, when the material evaluation is carried out, the user behavior data such as the sending time and the number of bullet screens, likes, forwards and other interactive behaviors, and the content recommendation algorithm of the platform are analyzed, and in terms of the advertising type, the Vlog type advertisement usually pays attention to the natural implantation of the content and the personal style of the creator, and the evaluation type advertisement emphasizes the performance display of the product and the trust degree of the audience. In this step, these data are collected and analyzed to provide a basis for subsequent advertising placement strategy, to ensure that the advertising content and the placement method match the characteristics of the platform and the advertising type, so as to improve the propagation effect and user acceptance of the advertisement.

[0157] S62: According to the advertising type and the placement platform, the material evaluation result is corrected again, and the corrected evaluation result is taken as the placement effect analysis result.

[0158] Specifically, after the characteristics of the advertising type and the placement platform are obtained, the preliminary material evaluation result is corrected again, and this process combines the original evaluation result with the platform user behavior mode and the advertising interactive data by deeply analyzing the characteristics of the advertising type and the placement platform, for example, in the Vlog type advertisement, the platform users tend to natural and personalized content, so the score of the hard advertisement is reduced, and the score of the content native advertisement is increased; in the evaluation type advertisement, more attention is paid to the clarity and credibility of product information to ensure that the placement effect of the advertisement on the specific platform reaches the best, in addition, for the streaming media platform, the interactive behavior data of the user such as the bullet screen and the like are focused on, and in the traditional media, more attention is paid to the advertising broadcast time period and the viewing habit of the audience. Through this series of correction, the final evaluation result will be more accurate, which can better guide the subsequent advertising placement and effect evaluation to ensure that the advertisement achieves the expected propagation effect.

[0159] Embodiment 2

[0160] See Figure 8 The embodiment 2 of the present application also provides an advertisement material effect evaluation device based on weighted fusion analysis, the device comprises:

[0161] A material uploading module is configured to obtain an uploaded original video material;

[0162] A material cutting module is configured to perform scene switching detection on the original video material, and cut the original video material into a plurality of scene video segments according to the scene switching detection result;

[0163] The blind watermark embedding module is configured to embed preset blind watermark information into each frame of image in each scene video segment by using an image processing technology based on a frequency domain transform.

[0164] The blind watermark extraction module is configured to extract blind watermark information from a target video highlight obtained by cutting each scene video segment in which the blind watermark information is embedded, and determine material segment information in the target video highlight according to the extracted blind watermark information.

[0165] The material evaluation module is configured to evaluate each material segment according to the pre-collected delivery data information and the material segment information when the target video highlight is delivered, and output a material evaluation result.

[0166] Specifically, the advertisement material effect evaluation device based on weighted fusion analysis provided by the embodiment of the application comprises: a material uploading module configured to obtain uploaded original video materials; a material cutting module configured to perform scene switching detection on the original video materials, and cut the original video materials into a plurality of scene video segments according to the scene switching detection result; a blind watermark embedding module configured to embed preset blind watermark information into each frame of image in each scene video segment by using an image processing technology based on a frequency domain transform; a blind watermark extraction module configured to extract blind watermark information from a target video highlight obtained by cutting each scene video segment in which the blind watermark information is embedded, and determine material segment information in the target video highlight according to the extracted blind watermark information; and a material evaluation module configured to evaluate each material segment according to the pre-collected delivery data information and the material segment information when the target video highlight is delivered, and output a material evaluation result. The device cuts the original video materials into a plurality of scene video segments by performing scene switching detection on the original video materials, and embeds blind watermark information into each frame of image in each scene segment by using an image processing technology based on a frequency domain transform, so that each material segment can be uniquely identified. In the target video highlight obtained by cutting, each original video material is accurately identified and tracked in position and use by extracting the blind watermark information. When the advertisement is delivered, the delivery effect of each original video material is accurately evaluated according to the pre-collected delivery data information and the extracted material segment information. The device not only solves the problem of weak correlation between the advertisement and the original video material, but also provides accurate material performance data through fine evaluation, which helps the advertiser to optimize the use of materials and improve the efficiency and effect of advertisement delivery. The systematic and automatic processing and analysis process reduces manual intervention and subjective bias, and improves the accuracy and real-time performance of the evaluation.

[0167] Embodiment 3

[0168] In addition, in combination with Figure 1The advertising material effect evaluation method based on weighted fusion analysis of the described embodiment 1 of the present application can be realized by an electronic device. Figure 9 A hardware structure schematic diagram of an electronic device provided by the embodiment 3 of the present application is shown.

[0169] The electronic device can include a processor and a memory having computer program instructions stored therein.

[0170] Specifically, the processor can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0171] The memory can include a mass storage for data or instructions. By way of example and not limitation, the memory can include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. Where appropriate, the memory can include removable or non-removable (or fixed) media. Where appropriate, the memory can be internal or external to the data processing device. In certain embodiments, the memory is a nonvolatile solid-state memory. In certain embodiments, the memory includes read-only memory (ROM). Where appropriate, this ROM can be mask-programmed ROM, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), electrically alterable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0172] The processor realizes any one of the advertising material effect evaluation methods based on weighted fusion analysis in the above embodiments by reading and executing the computer program instructions stored in the memory.

[0173] In one example, the electronic device can further include a communication interface and a bus. Wherein, as shown in Figure 9 The processor, the memory, the communication interface are connected through the bus and complete the communication among each other.

[0174] The communication interface is mainly used to realize the communication among the modules, devices, units and / or equipment in the embodiments of the present application.

[0175] The bus includes hardware, software, or both, coupling components of the device to each other and / or to input and output devices. As an example and not by way of limitation, the bus can include an accelerated graphics port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a front-side bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect (IB), a low pin count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association local (VLB) bus, or another suitable bus or combination thereof. As appropriate, the bus can include one or more buses. Although the present embodiments describe and show a particular bus, the present embodiments contemplate any suitable bus or interconnect.

[0176] The electronic device provided by the embodiment detects scene switching of the original video material, cuts the original video material into multiple scene video clips, and embeds blind watermark information into each frame image of each scene clip by using image processing technology based on frequency domain transformation, so that each material clip can be uniquely identified. In the target video highlight after editing, the position and use of each original video material are accurately identified and tracked by extracting the blind watermark information. When advertising is put, the advertising effect of each original video material is accurately evaluated according to the collected advertising data information and the extracted material clip information. Not only the problem of weak correlation between advertising and original video material is solved, but also accurate material performance data is provided through fine evaluation, which helps the advertiser to optimize the use of materials and improve the efficiency and effect of advertising. This systematic and automatic processing and analysis process reduces manual intervention and subjective bias, and improves the accuracy and real-time performance of the evaluation.

[0177] Embodiment 4

[0178] In addition, in combination with the advertising material effect evaluation method based on weighted fusion analysis in the above embodiment 1, the embodiment 4 can also provide a computer readable storage medium to realize. The computer readable storage medium has computer program instructions stored thereon; the computer program instructions are executed by the processor to realize any one of the advertising material effect evaluation methods based on weighted fusion analysis in the above embodiments.

[0179] The storage medium provided by the embodiment provides scene switching detection on the original video material, cuts the original video material into a plurality of scene video segments, and embeds blind watermark information in each frame image of each scene segment by using the image processing technology based on frequency domain transformation, so that each material segment can be uniquely identified. In the target video highlight after editing, the position and use of each original video material are accurately identified and tracked by extracting the blind watermark information. When the advertisement is put into operation, the delivery effect of each original video material is accurately evaluated according to the pre-collected delivery data information and the extracted material segment information. Not only the problem of weak correlation between the advertisement and the original video material is solved, but also accurate material performance data is provided through fine evaluation, which helps the advertiser to optimize the use of materials and improve the efficiency and effect of advertisement delivery. The systematic and automatic processing and analysis process reduces the manual intervention and subjective bias, and improves the accuracy and real-time performance of the evaluation.

[0180] In summary, the embodiment of the present application provides an advertisement material effect evaluation method, device and equipment based on weighted fusion analysis.

[0181] It should be noted that the present application is not limited to the specific configurations and processes described above and shown in the drawings. For the sake of brevity, detailed descriptions of known methods are omitted herein. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order of the steps, after understanding the spirit of the present application.

[0182] The user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of the place of residence, and provide corresponding operation portal for user to choose authorization or refusal.

[0183] It should also be noted that the exemplary embodiments mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be executed in the order mentioned in the embodiments, or different from the order in the embodiments, or several steps are executed simultaneously.

[0184] The above merely describes specific implementation of the present application, and those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, module and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be described herein again. It should be understood that the protection scope of the present application is not limited to this, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application.

Claims

1. An advertisement material effect evaluation method based on weighted fusion analysis, characterized by, The method comprises: Scene switching detection and blind watermark information embedding are performed on at least one original video material in an advertisement delivery scenario to obtain a target video highlight; Blind watermark information is extracted from the target video highlight, and material segment information is determined according to the extracted blind watermark information, wherein the material segment information at least includes a name and a position of the original video material used in the target video highlight; According to the use frequency, when the original video material is used multiple times, a single material evaluation result corresponding to each time of the original video material is calculated by weighting and averaging according to the real-time material play number to determine a multiple material evaluation result. After the multiple material evaluation result is determined according to the use frequency, when the original video material is used multiple times, the advertisement type and the delivery platform when the target video highlight is delivered are obtained; According to the advertisement type and the delivery platform, the material evaluation result is corrected again, and the corrected evaluation result is taken as a delivery effect analysis result; The single material evaluation result is determined by the following steps: The total duration of the target video highlight is obtained, and a material position gear is determined according to the original video material position and the total duration, wherein the material position gear is used to identify the relative time interval of the original video material in the target video highlight; According to the material position gear, a standard parameter is determined, wherein the standard parameter at least includes a standard material play number, a standard material click rate and a standard material loss rate; The standard parameter and the evaluation parameter are calculated to determine the single material evaluation result. The method comprises: The original video material is decomposed into multiple frames of original video images, and a color histogram corresponding to each original video image is obtained; 2. The advertisement material effect evaluation method based on weighted fusion analysis according to claim 1, characterized in that, Difference calculation is performed on the color histograms of adjacent frames of original video images to obtain a color histogram difference value; If the color histogram difference value is greater than a preset color histogram threshold value, the adjacent frames of original video images are classified into two scene video segments; By using an image processing technology based on frequency domain transformation, preset blind watermark information is embedded into each frame of image in each scene video segment, and each scene video segment after embedding the blind watermark information is edited to obtain a target video highlight. The method comprises: ​ 3. The advertisement material effect evaluation method based on weighted fusion analysis according to claim 2, characterized in that, ​ Converting each frame image in each scene video segment into a target color channel by color space conversion; Converting each frame image in each scene video segment into a scene frequency domain image by discrete cosine transform on the target color channel; Modifying a high-frequency region in the scene frequency domain image according to a preset key to embed preset blind watermark information into the scene frequency domain image.

4. The advertisement material effect evaluation method based on weighted fusion analysis according to claim 2, characterized in that, The preset color histogram threshold is determined by the following steps: Obtaining a current frame image in adjacent frame original video images; Determining a color histogram set of all frame images before the current frame image according to the current frame image; Determining the color histogram threshold according to the color histogram set, in combination with a preset smoothing parameter and an initial threshold.

5. The advertisement material effect evaluation method based on weighted fusion analysis according to claim 3, characterized in that, The modification processing on the high-frequency region in the scene frequency domain image according to the preset key to embed the preset blind watermark information into the scene frequency domain image includes: Obtaining a template size of a to-be-embedded region according to the preset blind watermark information; Selecting an arbitrary target region in the scene frequency domain image according to the template size of the to-be-embedded region, wherein the target region has the same size as the template size of the to-be-embedded region; Performing frequency domain coefficient analysis on the target region to calculate a sum of squares of all frequency domain coefficients in the target region as a region energy value; Comparing the region energy value with a preset energy threshold, and if the region energy value is greater than the energy threshold, classifying the target region as a high-frequency region; If the region energy value is less than or equal to the energy threshold, classifying the target region as a low-frequency region; Repeating the selection of the arbitrary target region in the scene frequency domain image for the frequency domain coefficient analysis until all regions in the scene frequency domain image are classified as the high-frequency region or the low-frequency region; Obtaining multiple scene frequency domain images, and selecting a scene frequency domain image as a current frame image; Analyzing positions of each pixel point in the high-frequency region in the current frame image and positions of each pixel point in a previous frame image by using an optical flow method to determine an optical flow field corresponding to each pixel point; Performing mean value calculation on the optical flow field corresponding to each pixel point to determine an optical flow mean value corresponding to the high-frequency region in the current frame image; When the optical flow mean value is less than a preset optical flow threshold, modifying the high-frequency region in the current frame image to embed the preset blind watermark information into the scene frequency domain image; Repeating the selection of the scene frequency domain image as the current frame image for the blind watermark information embedding until the preset blind watermark information is embedded into all scene frequency domain images.

6. A device for evaluating the effectiveness of advertising creatives based on weighted fusion analysis, characterized in that, The device includes: A target video highlight determination module configured to perform scene switching detection and blind watermark information embedding on at least one original video material in an advertisement delivery scene to obtain a target video highlight; A blind watermark information extraction module configured to extract blind watermark information from the target video highlight and determine material segment information according to the extracted blind watermark information, wherein the material segment information at least includes a name and a position of an original video material used in the target video highlight. The delivery data statistics module is configured to perform second-level statistics on the pre-collected delivery data according to the material segment information, and determine real-time evaluation parameters related to the original video material evaluation and the number of uses of the original video material in advertisement delivery, wherein the real-time evaluation parameters at least include real-time material play number, real-time material click rate and real-time material churn rate. The multiple-use material evaluation module is configured to, when the original video material is used multiple times according to the number of uses, perform weighted average calculation on single-material evaluation results corresponding to each original video material according to the real-time material play number, and determine a multiple-material evaluation result. After the multiple-material evaluation result is determined by performing weighted average calculation on single-material evaluation results corresponding to each original video material according to the real-time material play number when the original video material is used multiple times according to the number of uses, the method further includes: obtaining an advertisement type and a delivery platform when the target video highlight is used for advertisement delivery; performing secondary correction on the material evaluation result according to the advertisement type and the delivery platform, and taking the corrected evaluation result as a delivery effect analysis result; The single-material evaluation result is determined by the following steps: obtaining a total duration of the target video highlight, and determining a material position gear according to the original video material position and the total duration, wherein the material position gear is used to identify a relative time interval of the original video material in the target video highlight; determining standard parameters according to the material position gear, wherein the standard parameters at least include standard material play number, standard material click rate and standard material churn rate; calculating the standard parameters and the evaluation parameters to determine the single-material evaluation result.

7. An electronic device, comprising: The computer program instructions, when executed by the processor, implement the method of any one of claims 1-5. The computer program instructions, when executed by the processor, implement the method of any one of claims 1-5.

8. A storage medium having stored thereon computer program instructions, characterized in that, ​

Images