A ship video on demand method and system based on video slicing

Abstract

The present invention belongs to the field of ship communication, and discloses a ship video on-demand method and system based on video slicing, which obtains corresponding feature video slices by using the method of dimensional video slicing, marks the feature video slices with corresponding label features, and constructs a video slicing ship on-demand model by using the feature video slices, label features and corresponding user account information. The model processes the user account information to be verified to obtain the video slices to be on-demand and the corresponding complete video required by the user, and the complete video is transmitted to the user account on-demand terminal by the method of dimensional video slicing, and the user account on-demand terminal plays the pushed complete video on-demand, thereby realizing the sliced ​​transmission and personalized push of the on-demand video. The present invention improves the on-demand transmission rate of ship videos by constructing a video slicing ship on-demand model, and realizes accurate personalized push to users by using the localized video already available on the ship, effectively improving the entertainment quality of users on the ship.

Description

Technical Field

[0001] The present invention belongs to the field of ship communications, and in particular relates to a ship video-on-demand method and system based on video slicing. Background Art

[0002] The marine industry has an increasing demand for information services such as massive data transmission and online broadband entertainment. The marine video-on-demand system provides video-on-demand services for marine users.

[0003] On the one hand, the video source acquisition and update methods of the marine video on demand system are mainly through mobile network updates and mobile hard disk updates, but these two methods have obvious disadvantages. Updating through the mobile network means that the video source is updated through the shore-based mobile network when the ship is docked. This method cannot guarantee the freshness of the video. When the ship is far away from the coast, due to the unstable communication quality at sea, this method cannot normally acquire and update the video source. Most of the existing technologies use mobile hard disks to update video sources, which is too expensive. At the same time, the update cycle of the video source is long and there are great safety risks.

[0004] On the other hand, since video data is generally very large files, the transmission of very large data on ships can generally be done by slicing very large files or large files, including idle transmission, high data compression, data encryption, etc. However, in recent years, video slicing technology has been mostly used in classroom teaching videos, video entertainment editing and other fields, and has rarely been used in scenarios where the network on ships is poor.

[0005] In addition, if the existing technology only loads the video step by step for video on demand on ships, users may not find the desired video and waste time waiting for loading. The existing technology for video recommendation does not take into account the transmission of large video data in such extreme environments.

[0006] Through the above analysis, the existing technology has the following problems to be solved:

[0007] How to use video slicing technology to accurately recommend large entertainment video files in a ship environment, how to transmit the complete video to the user account on-demand terminal through dimensional video slicing, and the user account on-demand terminal automatically recommends the pushed complete video for on-demand playback, and realizes sliced ​​transmission and personalized push of on-demand video is an urgent problem to be solved in the field of ship video transmission.

[0008] Therefore, it is necessary to design a ship video on demand method and system based on video slicing to address the existing problems. Summary of the invention

[0009] In order to solve the above technical problems, the present invention proposes a ship video on demand method and system based on video slicing.

[0010] In a first aspect of the present invention, a ship video on demand method based on video slicing is proposed;

[0011] Acquire a local video of the ship, process the local video of the ship to obtain a first feature video slice, and label the first feature video slice to obtain a first label feature;

[0012] Obtain the user's first account information, and obtain the first recommended video set corresponding to this time;

[0013] A video slice ship on-demand model is trained according to the first feature video slice, the first label feature, the first account information, and the first recommended video, where the video slice ship on-demand model is a deep learning model;

[0014] The second feature video slice and the second label feature are obtained, and the second account information is obtained. The video slice ship on-demand model processes the second feature video slice, the second label feature and the second account information to obtain a second recommended video. The second recommended video is used to push the ship video on-demand to the user.

[0015] Furthermore, the first characteristic video slice or the second characteristic video slice is obtained by using a video slice extraction method, wherein the video slice extraction method extracts the local video of the ship as containing two-dimensional spatial information (P X , P y ) and the time feature t, the first feature video slice or the second feature video slice F is:

[0016] ;

[0017] In the formula Indicates the two-dimensional spatial information corresponding to time i (P X , P y ) Brightness feature value of the image, two-dimensional spatial information P X and P y The position parameters of the slice are expressed in the two-dimensional coordinates of the space constructed by the slice.

[0018] Furthermore, the brightness characteristic value of the two-dimensional spatial information corresponding to the time i is The calculation method is as follows:

[0019] ;

[0020] In the formula, , , is a constant value determined according to the ship video client standard. , and is the color channel value of the slice image at time i.

[0021] Furthermore, the video slice ship on-demand model is constructed using a classifier model based on the Fisher criterion or an improved CNN model.

[0022] Furthermore, an improved CNN model is adopted, and its activation function is:

[0023] ;

[0024] is the activation function value, , , is a constant value determined according to the ship video client standard, F is the first feature video slice or the second feature video slice, and C is the input feature value.

[0025] A ship video on demand system based on video slicing is also provided, which includes a ship video slicing module, a video tag module, a ship client module, a video slicing on demand model building module and a video slicing on demand module:

[0026] The ship video slicing module processes the acquired local ship video to obtain a first feature video slice, and the video labeling module labels the first feature video slice to obtain a first label feature;

[0027] The ship client module obtains the first account information of the user and obtains the first recommended video corresponding to the setting at this time;

[0028] The video slice on-demand model construction module trains a video slice ship on-demand model according to the first feature video slice, the first label feature, the first account information and the first recommended video, and the video slice ship on-demand model is a deep learning model;

[0029] The video slice on-demand module obtains the second characteristic video slice and the second label feature, and obtains the second account information. The video slice on-demand module calls the video slice ship on-demand model to process the second characteristic video slice, the second label feature and the second account information to obtain a second recommended video. The video slice on-demand module uses the second recommended video to push the ship video on-demand to the user.

[0030] Furthermore, the first characteristic video slice or the second characteristic video slice is obtained by using a video slice extraction method, wherein the video slice extraction method extracts the local video of the ship as containing two-dimensional spatial information (P X , Py ) and the time feature t, the first feature video slice or the second feature video slice F is:

[0031] ;

[0032] In the formula Indicates the two-dimensional spatial information corresponding to time i (P X , P y ) Brightness feature value of the image, two-dimensional spatial information P X and P y The position parameters of the slice are expressed in the two-dimensional coordinates of the space constructed by the slice.

[0033] Furthermore, the brightness characteristic value of the two-dimensional spatial information corresponding to the time i is The calculation method is as follows:

[0034] ;

[0035] In the formula, , , is a constant value determined according to the ship video client standard. , and is the color channel value of the slice image at time i.

[0036] Furthermore, the video slice ship on-demand model is constructed using a classifier model based on the Fisher criterion or an improved CNN model.

[0037] Furthermore, an improved CNN model is adopted, and its activation function is:

[0038] ;

[0039] is the activation function value, , , is a constant value determined according to the ship video client standard, F is the first feature video slice or the second feature video slice, and C is the input feature value.

[0040] The present invention provides a ship video on-demand method and system based on video slicing. The video slicing technology is used to convert the video image into a light feature value for label slicing correspondence, thereby constructing a big data neural network to improve the CNN model to accurately recommend the ultra-large entertainment video files in the ship environment. The present invention transmits the complete video to the user account on-demand terminal through the dimensional video slicing method, and the user account on-demand terminal automatically recommends the pushed complete video for on-demand playback, thereby realizing the sliced ​​transmission and personalized push of the on-demand video. In addition, the present invention does not require the user to wait for a long time to load. The video recommendation technology takes into account the transmission of ultra-large video data in extreme environments, realizes accurate push, rather than the comprehensive loading of existing videos, improves the on-demand transmission rate of ship videos, and uses the existing localized videos of the ship to realize accurate personalized push to users, effectively improving the entertainment quality of users on the ship.

[0041] More embodiments and improved effects of the present invention will be further introduced in conjunction with the drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flow chart of a ship video on demand method based on video slicing of the present invention;

[0043] Figure 2 It is a schematic diagram of a ship video-on-demand system based on video slicing of the present invention;

[0044] Figure 3 is a schematic diagram of a video on-demand recommendation based on the system in an embodiment of the present invention;

[0045] Figure 4 It is a schematic diagram of the convolution principle after improving the CNN model activation function in the embodiment of the present invention;

[0046] Figure 5 It is a schematic diagram of the structure of an electronic device for implementing the method of the present invention according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The invention is further described below in conjunction with the accompanying drawings and specific implementation methods.

[0048] In order to solve the above technical problems, the present invention proposes a ship video on demand method and system based on video slicing.

[0049] In a first aspect of the present invention, a ship video on demand method based on video slicing is proposed;

[0050] Acquire a local video of the ship, process the local video of the ship to obtain a first feature video slice, and label the first feature video slice to obtain a first label feature;

[0051] In this embodiment, the label features can be divided into artificially set labels such as science fiction, thriller, variety show, etc., and the label values ​​can be listed as 0, 1, 2, etc.

[0052] Obtain the user's first account information, and obtain the first recommended video set corresponding to this time;

[0053] A video slice ship on-demand model is trained according to the first feature video slice, the first label feature, the first account information, and the first recommended video, where the video slice ship on-demand model is a deep learning model;

[0054] The second feature video slice and the second label feature are obtained, and the second account information is obtained. The video slice ship on-demand model processes the second feature video slice, the second label feature and the second account information to obtain a second recommended video. The second recommended video is used to push the ship video on-demand to the user.

[0055] In this embodiment, the first account information or the second account information is the user's age, gender, occupation, education level, etc., which are converted into corresponding feature values ​​as needed. For example, the feature value for age under 18 is 0, 18 to 30 years old is 1, and over 30 years old is -1. The gender is set to 1 for male and 0 for female. The rest are set according to actual needs.

[0056] Furthermore, the first characteristic video slice or the second characteristic video slice is obtained by using a video slice extraction method, wherein the video slice extraction method extracts the local video of the ship as containing two-dimensional spatial information (P X , P y ) and the time feature t, the first feature video slice or the second feature video slice F is:

[0057] ;

[0058] In the formula Indicates the two-dimensional spatial information corresponding to time i (P X , P y ) Brightness feature value of the image, two-dimensional spatial information P X and P y The position parameters of the slice are expressed in the two-dimensional coordinates of the space constructed by the slice.

[0059] According to this formula, in fact, the first feature video slice or the second feature video slice is obtained by calculating the average value of the entire brightness feature value of the video slice.

[0060] In this embodiment, the construction of the two-dimensional spatial information coordinate system is based on the Cartesian coordinate system, and the origin is selected as any point of the image.

[0061] Those skilled in the art know that in the field of digital image processing, a video is formed by quickly playing back pictures, and the slice images in this embodiment are pictures corresponding to a certain frame number at a certain moment in the video, that is, pictures of a certain frame number at a certain moment in the video are extracted as corresponding slice images.

[0062] Furthermore, the brightness characteristic value of the two-dimensional spatial information corresponding to the time i is The calculation method is as follows:

[0063] ;

[0064] In the formula, , , is a constant value determined according to the ship video client standard. , and is the color channel value of the slice image at time i.

[0065] In this embodiment , , They are 0.229, 0.707, 0.071 respectively. , and The range is 0 to 255.

[0066] Furthermore, the video slice ship on-demand model is constructed using a classifier model based on the Fisher criterion or an improved CNN model.

[0067] The CNN model in this embodiment conventionally has an input layer, a convolution layer, a pooling layer, an activation function and a fully connected layer and an output layer. In this embodiment, the activation function is modified and changed based on the parameters related to the brightness feature value for the corresponding application scenario, and the output classification recommendation video obtained for the ship video slice is more accurate.

[0068] Furthermore, an improved CNN model is adopted, and its activation function is:

[0069] ;

[0070] is the activation function value, , , is a constant value determined according to the ship video client standard, F is the first feature video slice or the second feature video slice, and C is the input feature value.

[0071] A ship video on demand system based on video slicing is also provided, which includes a ship video slicing module, a video tag module, a ship client module, a video slicing on demand model building module and a video slicing on demand module:

[0072] The ship video slicing module processes the acquired local ship video to obtain a first feature video slice, and the video labeling module labels the first feature video slice to obtain a first label feature;

[0073] The ship client module obtains the first account information of the user and obtains the first recommended video corresponding to the setting at this time;

[0074] The video slice on-demand model construction module trains a video slice ship on-demand model according to the first feature video slice, the first label feature, the first account information and the first recommended video, and the video slice ship on-demand model is a deep learning model;

[0075] The video slice on-demand module obtains the second characteristic video slice and the second label feature, and obtains the second account information. The video slice on-demand module calls the video slice ship on-demand model to process the second characteristic video slice, the second label feature and the second account information to obtain a second recommended video. The video slice on-demand module uses the second recommended video to push the ship video on-demand to the user.

[0076] Furthermore, the first characteristic video slice or the second characteristic video slice is obtained by using a video slice extraction method, wherein the video slice extraction method extracts the local video of the ship as containing two-dimensional spatial information (P X , P y ) and the time feature t, the first feature video slice or the second feature video slice F is:

[0077] ;

[0078] In the formula Indicates the two-dimensional spatial information corresponding to time i (P X , P y ) Brightness feature value of the image, two-dimensional spatial information P X and P y The position parameters of the slice are expressed in the two-dimensional coordinates of the space constructed by the slice.

[0079] According to this formula, in fact, the first feature video slice or the second feature video slice is obtained by calculating the average value of the entire brightness feature value of the video slice.

[0080] Furthermore, the brightness characteristic value of the two-dimensional spatial information corresponding to the time i is The calculation method is as follows:

[0081] ;

[0082] In the formula, , , is a constant value determined according to the ship video client standard. , and is the color channel value of the slice image at time i.

[0083] Furthermore, the video slice ship on-demand model is constructed using a classifier model based on the Fisher criterion or an improved CNN model.

[0084] Furthermore, an improved CNN model is adopted, and its activation function is:

[0085] ;

[0086] is the activation function value, , , is a constant value determined according to the ship video client standard, F is the first feature video slice or the second feature video slice, and C is the input feature value.

[0087] The present invention provides a ship video on-demand method and system based on video slicing. The video slicing technology is used to convert the video image into a light feature value for label slicing correspondence, thereby constructing a big data neural network to improve the CNN model to accurately recommend super-large entertainment video files in the ship environment. The present invention transmits the complete video to the user account on-demand terminal through the dimensional video slicing method, and the user account on-demand terminal plays the pushed complete video on-demand, thereby realizing the sliced ​​transmission and personalized push of the on-demand video. In addition, the present invention does not require the user to wait for a long time to load. The video recommendation technology takes into account the transmission of super-large video data in extreme environments, realizes accurate push, rather than the comprehensive loading of existing videos, improves the on-demand transmission rate of ship videos, and uses the existing localized videos of the ship to realize accurate personalized push to users, effectively improving the entertainment quality of users on the ship.

[0088] Of course, it can be understood that each embodiment of the present invention can achieve one of the effects alone, and a combination of multiple embodiments of the present invention can achieve all of the above effects, but it is not required that every embodiment of the present invention achieve all of the above advantages and effects, because each embodiment of the present invention can constitute a separate technical solution and make one or more contributions to the prior art.

[0089] The present invention does not particularly specify the structure of some modules, which shall be subject to the contents recorded in the prior art. The prior art mentioned in the above background technology section and the specific embodiment section of the present invention can be used as part of the present invention to understand the meaning of some technical features or parameters. The scope of protection of the present invention shall be subject to the contents actually recorded in the claims.

Claims

1. A ship video on demand method based on video slicing, characterized in that: Acquire a local video of the ship, process the local video of the ship to obtain a first feature video slice, and label the first feature video slice to obtain a first label feature; Obtain the user's first account information, and obtain the first recommended video set corresponding to this time; A video slice ship on-demand model is trained according to the first feature video slice, the first label feature, the first account information, and the first recommended video, where the video slice ship on-demand model is a deep learning model; A second feature video slice and a second label feature are obtained, and second account information is obtained. The video slice ship on-demand model processes the second feature video slice, the second label feature and the second account information to obtain a second recommended video. The second recommended video is used to push the ship video on-demand to the user. The first characteristic video slice or the second characteristic video slice is obtained by using a video slice extraction method, wherein the video slice extraction method extracts the local video of the ship as containing two-dimensional spatial information (P X , P y ) and the time feature t, the first feature video slice or the second feature video slice F is: ； In the formula Indicates the two-dimensional spatial information corresponding to time i (P X , P y ) Brightness feature value of the image, two-dimensional spatial information P X and P y is the slice position parameter expressed by the two-dimensional coordinates of the space constructed by the slice; The moment i corresponds to the brightness characteristic value of the two-dimensional spatial information The calculation method is as follows: ； In the formula, , , is a constant value determined according to the ship video client standard. , and is the color channel value of the slice image at time i; The video slicing ship on-demand model is constructed using a classifier model based on the Fisher criterion or an improved CNN model.

2. A ship video on demand method based on video slicing as claimed in claim 1, characterized in that: The improved CNN model is adopted, and its activation function is: ； is the activation function value, , , is a constant value determined according to the ship video client standard, F is the first feature video slice or the second feature video slice, and C is the input feature value.

3. A ship video on demand system based on video slicing, characterized in that: The ship video on demand system includes a ship video slicing module, a video tag module, a ship client module, a video slicing on demand model building module and a video slicing on demand module: The ship video slicing module processes the acquired local ship video to obtain a first feature video slice, and the video labeling module labels the first feature video slice to obtain a first label feature; The ship client module obtains the first account information of the user and obtains the first recommended video corresponding to the setting at this time; The video slice on-demand model construction module trains a video slice ship on-demand model according to the first feature video slice, the first label feature, the first account information and the first recommended video, and the video slice ship on-demand model is a deep learning model; The video slice on-demand module obtains a second characteristic video slice and a second label feature, and obtains second account information. The video slice on-demand module calls the video slice ship on-demand model to process the second characteristic video slice, the second label feature and the second account information to obtain a second recommended video. The video slice on-demand module uses the second recommended video to push the ship video on-demand to the user. The first characteristic video slice or the second characteristic video slice is obtained by using a video slice extraction method, wherein the video slice extraction method extracts the local video of the ship as containing two-dimensional spatial information (P X , P y ) and the time feature t, the first feature video slice or the second feature video slice F is: ； In the formula Indicates the two-dimensional spatial information corresponding to time i (P X , P y ) Brightness feature value of the image, two-dimensional spatial information P X and P y is the slice position parameter expressed by the two-dimensional coordinates of the space constructed by the slice; The moment i corresponds to the brightness characteristic value of the two-dimensional spatial information The calculation method is as follows: ； In the formula, , , is a constant value determined according to the ship video client standard. , and is the color channel value of the slice image at time i; The video slicing ship on-demand model is constructed using a classifier model based on the Fisher criterion or an improved CNN model.

4. A ship video on demand system based on video slicing as claimed in claim 3, characterized in that: The improved CNN model is adopted, and its activation function is: ； is the activation function value, , , is a constant value determined according to the ship video client standard, F is the first feature video slice or the second feature video slice, and C is the input feature value.

Images