Method, device, equipment and storage medium for detecting quality of ring back tone video

Abstract

The present invention discloses a method, apparatus, device, and computer-readable storage medium for detecting the quality of a ringback tone video. The method comprises: obtaining a screen recording corresponding to a played ringback tone video; determining a first pixel difference between the source video of the ringback tone video and the screen recording; and determining whether the ringback tone video is abnormal based on the first pixel difference. The present invention improves the accuracy of abnormal ringback tone video detection.

Description

[0001] Technology Neighborhood

[0002] The present invention relates to the field of computer technology, and in particular to a method, device, equipment and computer-readable storage medium for detecting the quality of a color ring tone video. Background Art

[0003] Video ringback tone (RBT) is a new service experience that includes short video content. When a user uses a terminal that supports RBT to initiate an audio or video call to a user with RBT over a VoLTE (Voice over Long-Term Evolution) network, the system plays a video clip to the calling user, improving the call waiting experience.

[0004] However, the existing technology can only detect whether the video ringback tone is played successfully based on the response value returned by the terminal, but cannot detect whether quality abnormalities occur during the video ringback tone playback. Summary of the Invention

[0005] The main purpose of the present invention is to provide a method, device, equipment and computer-readable storage medium for detecting the quality of a color ring tone video, aiming to solve the problem of how to detect whether quality abnormalities occur during the playback of a video color ring tone.

[0006] To achieve the above object, the present invention provides a method for detecting the quality of a ring back tone video, the method comprising the following steps:

[0007] Get the screen recording video corresponding to the played ringback tone video;

[0008] Determining a first pixel difference between a source video of a ring back tone video and the screen recording video;

[0009] Determine whether the CRBT video is an abnormal CRBT video according to the first pixel difference.

[0010] In one embodiment, the step of determining a first pixel difference between the source video of the ringback tone video and the screen recording video includes:

[0011] Determine a first pixel difference between a video frame of a source video of the ring back tone video and a corresponding video frame of the screen recording video, where the first pixel difference includes pixel differences corresponding to each video frame.

[0012] In one embodiment, the step of determining whether the CRBT video is an abnormal CRBT video according to the first pixel difference includes:

[0013] Inputting the first pixel difference into a preset quality detection model to obtain a second pixel difference output by the quality detection model, wherein the quality detection model is trained by normal video;

[0014] determining a reconstruction error according to the first pixel difference and the second pixel difference;

[0015] If the reconstruction error is greater than a preset threshold, determining that the CRBT video is an abnormal CRBT video;

[0016] If the reconstruction error is less than or equal to a preset threshold, it is determined that the CRBT video is a normal CRBT video.

[0017] In one embodiment, the step of determining the reconstruction error according to the first pixel difference and the second pixel difference includes:

[0018] determining a Euclidean distance between the first pixel difference and the second pixel difference;

[0019] The reconstruction error is determined according to the Euclidean distance.

[0020] In one embodiment, before the step of inputting the first pixel difference into a preset quality detection model, the method further includes:

[0021] Obtaining a training set and a test set, wherein the training set includes normal videos, and the test set includes normal videos and abnormal videos;

[0022] Training a preset neural network model according to the training set, wherein the output of the neural network model is a second pixel difference of a normal video in the training set;

[0023] Determining a training error of the trained neural network model based on the test set;

[0024] When the training error is less than a preset error value, the trained neural network model is used as a quality detection model.

[0025] In one embodiment, before the step of obtaining the training set and the test set, the method further includes:

[0026] Obtain historical ringback tone videos and abnormal ringback tone videos;

[0027] Determine the pixel mean of the pixel points at the same position of each video frame in the historical ring back tone video;

[0028] Subtract the pixel value of each pixel point in the historical ringback tone video from the pixel point average value corresponding to the pixel point at the same position to obtain a normal video;

[0029] Determine the pixel mean of the pixel points at the same position of each video frame in the abnormal ring back tone video;

[0030] The abnormal video is obtained by subtracting the pixel value of each video frame in the abnormal ring back tone video from the pixel mean value corresponding to the pixel at the same position.

[0031] In one embodiment, the step of obtaining the screen recording video corresponding to the played ringback tone video includes:

[0032] Get the source ringback tone video and screen recording ringback tone video;

[0033] Determine the pixel mean of the pixel points at the same position of each video frame in the source ring back tone video;

[0034] Subtract the pixel value of each pixel in the source ring back tone video from the pixel mean value of the pixel at the same position to obtain the source video;

[0035] Determine the pixel mean of the pixel points at the same position of each video frame in the screen recording ringback tone video;

[0036] The pixel value of each video frame in the screen recording ringback tone video is subtracted from the pixel mean value corresponding to the pixel at the same position to obtain the screen recording video.

[0037] To achieve the above object, the present invention further provides a device for detecting the quality of a ring back tone video, the device comprising:

[0038] The acquisition module is used to obtain the screen recording video corresponding to the played ringback tone video;

[0039] A determination module, configured to determine a first pixel difference between a source video of a ring back tone video and the screen recording video;

[0040] A detection module is configured to determine whether the CRBT video is an abnormal CRBT video according to the first pixel difference.

[0041] To achieve the above-mentioned objectives, the present invention also provides a quality detection device for a color ring back tone video, wherein the quality detection device for the color ring back tone video includes a memory, a processor, and a quality detection program for the color ring back tone video stored in the memory and executable on the processor. When the quality detection program for the color ring back tone video is executed by the processor, the various steps of the quality detection method for the color ring back tone video described above are implemented.

[0042] To achieve the above objectives, the present invention also provides a computer-readable storage medium, which stores a quality detection program for a ring back tone video. When the quality detection program for the ring back tone video is executed by a processor, the various steps of the above-mentioned method for quality detection of the ring back tone video are implemented.

[0043] The present invention provides a method, apparatus, device, and computer-readable storage medium for detecting the quality of a ringback tone video. These methods obtain a screen recording corresponding to a played ringback tone video, determine a first pixel difference between the source video and the screen recording, and determine whether the ringback tone video is abnormal based on the first pixel difference. By determining the first pixel difference between the screen recording and the source video, it is determined whether the ringback tone video differs significantly from the source video during playback. Videos with significant differences are identified as abnormal ringback tone videos, thereby improving the accuracy of abnormal ringback tone video detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 Schematic diagram of the hardware structure of a quality detection device for a color ring back tone video according to an embodiment of the present invention;

[0045] Figure 2 1. A flow chart of a first embodiment of a method for detecting the quality of a color ring back tone video according to the present invention;

[0046] Figure 3 This is a detailed flowchart of step S30 of the second embodiment of the method for detecting the quality of a color ring back tone video according to the present invention;

[0047] Figure 4 4 is a detailed flowchart of step S30 of the third embodiment of the method for detecting the quality of a color ring back tone video according to the present invention;

[0048] Figure 5 Schematic diagram of a quality detection model for a ring back tone video according to the present invention;

[0049] Figure 6 2 is a flow chart of a fourth embodiment of a method for detecting quality of a color ring back tone video according to the present invention;

[0050] Figure 7 The figure is a schematic diagram of the logic modules of the device for detecting the quality of the color ring tone video according to the present invention.

[0051] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0052] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0053] The main solution of the embodiment of the present invention is: obtaining the screen recording video corresponding to the played ringback tone video, determining the first pixel difference between the source video and the screen recording video of the ringback tone video, and determining whether the ringback tone video is an abnormal ringback tone video based on the first pixel difference.

[0054] By determining the first pixel difference between the recorded video and the source video, it is determined whether the ringback tone video has a large difference from the source video when playing, and the one with a large difference is determined to be an abnormal ringback tone video, thereby improving the accuracy of abnormal ringback tone video detection.

[0055] As an implementation solution, the quality detection device of the ring back tone video can be as follows Figure 1 shown.

[0056] The embodiment of the present invention relates to a quality detection device for a ring back tone video, which includes a processor 101, such as a CPU, a memory 102, and a communication bus 103. The communication bus 103 is used to achieve connection and communication between these components.

[0057] The memory 102 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. Figure 1 As shown, the memory 102 as a computer-readable storage medium may include a quality detection program for the CRBT video; and the processor 101 may be configured to call the quality detection program for the CRBT video stored in the memory 102 and perform the following operations:

[0058] Get the screen recording video corresponding to the played ringback tone video;

[0059] Determining a first pixel difference between a source video of a ring back tone video and the screen recording video;

[0060] Determine whether the CRBT video is an abnormal CRBT video according to the first pixel difference.

[0061] In one embodiment, the processor 101 may be configured to call a quality detection program for a color ring back tone video stored in the memory 102 and perform the following operations:

[0062] Determine a first pixel difference between a video frame of a source video of the ring back tone video and a corresponding video frame of the screen recording video, where the first pixel difference includes pixel differences corresponding to each video frame.

[0063] In one embodiment, the processor 101 may be configured to call a quality detection program for a color ring back tone video stored in the memory 102 and perform the following operations:

[0064] Inputting the first pixel difference into a preset quality detection model to obtain a second pixel difference output by the quality detection model, wherein the quality detection model is trained by normal video;

[0065] determining a reconstruction error according to the first pixel difference and the second pixel difference;

[0066] If the reconstruction error is greater than a preset threshold, determining that the CRBT video is an abnormal CRBT video;

[0067] If the reconstruction error is less than or equal to a preset threshold, it is determined that the CRBT video is a normal CRBT video.

[0068] In one embodiment, the processor 101 may be configured to call a quality detection program for a color ring back tone video stored in the memory 102 and perform the following operations:

[0069] determining a Euclidean distance between the first pixel difference and the second pixel difference;

[0070] The reconstruction error is determined according to the Euclidean distance.

[0071] In one embodiment, the processor 101 may be configured to call a quality detection program for a color ring back tone video stored in the memory 102 and perform the following operations:

[0072] Obtaining a training set and a test set, wherein the training set includes normal videos, and the test set includes normal videos and abnormal videos;

[0073] Training a preset neural network model according to the training set, wherein the output of the neural network model is a second pixel difference of a normal video in the training set;

[0074] Determining a training error of the trained neural network model based on the test set;

[0075] When the training error is less than a preset error value, the trained neural network model is used as a quality detection model.

[0076] In one embodiment, the processor 101 may be configured to call a quality detection program for a color ring back tone video stored in the memory 102 and perform the following operations:

[0077] Obtain historical ringback tone videos and abnormal ringback tone videos;

[0078] Determine the pixel mean of the pixel points at the same position of each video frame in the historical ring back tone video;

[0079] Subtract the pixel value of each pixel point in the historical ringback tone video from the pixel point average value corresponding to the pixel point at the same position to obtain a normal video;

[0080] Determine the pixel mean of the pixel points at the same position of each video frame in the abnormal ring back tone video;

[0081] The abnormal video is obtained by subtracting the pixel value of each video frame in the abnormal ring back tone video from the pixel mean value corresponding to the pixel at the same position.

[0082] In one embodiment, the processor 101 may be configured to call a quality detection program for a color ring back tone video stored in the memory 102 and perform the following operations:

[0083] Get the source ringback tone video and screen recording ringback tone video;

[0084] Determine the pixel mean of the pixel points at the same position of each video frame in the source ring back tone video;

[0085] Subtract the pixel value of each pixel in the source ring back tone video from the pixel mean value of the pixel at the same position to obtain the source video;

[0086] Determine the pixel mean of the pixel points at the same position of each video frame in the screen recording ringback tone video;

[0087] The pixel value of each video frame in the screen recording ringback tone video is subtracted from the pixel mean value corresponding to the pixel at the same position to obtain the screen recording video.

[0088] Based on the hardware architecture of the above-mentioned CRBT video quality detection device, an embodiment of the CRBT video quality detection method of the present invention is proposed.

[0089] Reference Figure 2 , Figure 2 This is a first embodiment of the method for detecting the quality of a ring back tone video of the present invention, and the method for detecting the quality of the ring back tone video includes the following steps:

[0090] Step S10, obtaining the screen recording video corresponding to the played ringback tone video.

[0091] Specifically, when the ring back tone video is played on the terminal device, the screen of the played ring back tone video is recorded to obtain a screen recording video.

[0092] Obtaining the screen recording video corresponding to the played ringtone video can be a terminal device obtaining the screen recording video corresponding to the played ringtone video for processing, where the terminal device can be an electronic device such as a mobile phone or tablet computer; or a server obtaining the screen recording video corresponding to the played ringtone video for processing, where the terminal device records the played ringtone video to obtain the screen recording video, and the server obtains the screen recording video sent by the terminal device.

[0093] Step S20: determining a first pixel difference between the source video of the ringback tone video and the screen recording video.

[0094] Specifically, the source video of the ring back tone video is sent by the server to the terminal device and played on the terminal device, wherein the video frames of the source video of the ring back tone video correspond to the video frames of the screen recording video one by one.

[0095] Determine the first pixel difference between the video frame of the source video of the ringback tone video and the video frame of the corresponding screen recording video, wherein the first pixel difference includes the pixel difference corresponding to each video frame. Determine the first pixel value of each first pixel point in each video frame of the source video, determine the second pixel value of each second pixel point in each video frame of the screen recording video, and determine the pixel difference between the first pixel value of each first pixel point of the source video and the second pixel value of the second pixel point of the corresponding screen recording video; determine the first pixel difference based on the pixel difference corresponding to each video frame. Exemplarily, the first pixel value of each first pixel point of the second video frame of the source video of the ringback tone video is represented by the following matrix:

[0096]

[0097] The second pixel value of each second pixel point of the second video frame of the screen recording video of the ringback tone video is represented by the following matrix:

[0098]

[0099] Each pixel point of the source video frame corresponds to each pixel point of the screen recording video frame in position. Calculate the pixel value of each pixel point of the second frame of the source video minus the absolute value of the pixel value of the second frame of the screen recording video at the same position to obtain the pixel difference corresponding to the second video frame, which is represented by the following matrix:

[0100]

[0101] By analogy, the pixel difference between each video frame of the source video and the corresponding video frame of the screen recording video can be obtained to obtain the first pixel difference.

[0102] Step S30: determining whether the CRBT video is an abnormal CRBT video according to the first pixel difference.

[0103] Specifically, whether the CRBT video is an abnormal CRBT video is determined based on the first pixel difference, wherein when there is a pixel difference greater than a preset threshold in the first pixel difference, the CRBT video is determined to be an abnormal CRBT video; for example, the preset threshold is 5, and when the pixel difference of the second video frame is If a pixel difference greater than 5 exists in the first pixel differences, the CRBT video is determined to be an abnormal CRBT video.

[0104] If the pixel difference in the first pixel difference does not exceed the preset threshold, the CRBT video is determined to be a normal CRBT video. For example, the preset threshold is 20, and the pixel difference in the second video frame is If there is no value greater than 20 in the matrix corresponding to the pixel difference, it is determined that the ringback tone video is a normal ringback tone video.

[0105] When the terminal device executes steps S10 to S30, after the terminal device determines that the CRBT video is an abnormal CRBT video, a prompt message is generated on the terminal device, and the recognition result of the abnormal CRBT video is returned to the server.

[0106] When the server executes steps S10 to S30, after determining that the CRBT video is an abnormal CRBT video, the server sends the recognition result or prompt information of the abnormal CRBT video to the terminal device.

[0107] In the technical solution of this embodiment, the screen recording corresponding to the played CRBT video is obtained, the first pixel difference between the source video and the screen recording is determined, and whether the CRBT video is an abnormal CRBT video is determined based on the first pixel difference. By determining the first pixel difference between the screen recording and the source video, it is determined whether the CRBT video is significantly different from the source video during playback. Videos with significant differences are identified as abnormal CRBT videos, thereby improving the accuracy of abnormal CRBT video detection.

[0108] Reference Figure 3 , Figure 3 This is a second embodiment of the method for detecting the quality of a ring back tone video according to the present invention. Based on the first embodiment, step S30 includes:

[0109] Step S31: inputting the first pixel difference into a preset quality detection model to obtain a second pixel difference output by the quality detection model, wherein the quality detection model is trained by normal video;

[0110] Step S32, determining a reconstruction error according to the first pixel difference and the second pixel difference;

[0111] Step S33: If the reconstruction error is greater than a preset threshold, determining that the CRBT video is an abnormal CRBT video;

[0112] Step S34: If the reconstruction error is less than or equal to a preset threshold, it is determined that the CRBT video is a normal CRBT video.

[0113] Specifically, the first pixel difference is input into a preset quality detection model to obtain a second pixel difference output by the quality detection model, wherein the quality detection model is trained by a source video of a normal video.

[0114] Determining the reconstruction error based on the first pixel difference and the second pixel difference may involve determining a Euclidean distance between the first pixel difference and the second pixel difference, and determining the reconstruction error based on the Euclidean distance, as shown in the following formula:

[0115] L(t)=||x(t)-x'(t)-fw(|x(t)-x'(t)|)||2;

[0116] Where x(t) represents the video frame sequence of the screen recording video, x'(t) represents the video frame sequence of the source video, x(t)-x'(t) represents the first pixel difference, and fw(|x(t)-x'(t)|) represents the second pixel difference output by the quality detection model.

[0117] Determining whether the ringback tone video is an abnormal ringback tone video based on a reconstruction error and a preset threshold, wherein if the reconstruction error is greater than the preset threshold, the ringback tone video is determined to be an abnormal ringback tone video, and if the reconstruction error is less than or equal to the preset threshold, the ringback tone video is determined to be a normal ringback tone video. The preset threshold can be manually determined; it can also be determined by the recall rate and precision rate of a quality detection model. The recall rate and precision rate of the quality detection model are determined, and a corresponding relationship between the recall rate and the precision rate is determined. A PRC (precision recall curve) is plotted with the recall rate and the precision rate as axes. When the precision rate in the PRC curve meets a preset range, the error value corresponding to the highest recall rate is determined as the preset threshold.

[0118] In the technical solution of this embodiment, the first pixel difference is input into a preset quality detection model, and the second pixel difference output by the quality detection model is obtained. A reconstruction error is determined based on the first and second pixel differences, and whether the RBT video is abnormal is determined based on the reconstruction error and a preset threshold. Determining the reconstruction error based on the first and second pixel differences and determining an abnormal RBT video based on the reconstruction error improves the accuracy of abnormal RBT video detection.

[0119] Reference Figure 4 , Figure 4 This is a third embodiment of the method for detecting the quality of a ring back tone video according to the present invention. Based on the second embodiment, before step S31, the method further includes:

[0120] Step S35, obtaining a training set and a test set, wherein the training set includes normal videos, and the test set includes normal videos and abnormal videos;

[0121] Step S36, training a preset neural network model according to the training set, wherein the output of the neural network model is the second pixel difference of the normal video in the training set;

[0122] Step S37, determining the training error of the trained neural network model based on the test set;

[0123] Step S38: When the training error is less than a preset error value, the trained neural network model is used as a quality detection model.

[0124] Specifically, a training set and a test set are obtained, wherein the training set includes normal videos, and the test set includes normal videos and abnormal videos.

[0125] Before obtaining the training set and the test set, the historical ringback tone videos are preprocessed to obtain normal videos, and the abnormal ringback tone videos are preprocessed to obtain abnormal videos.

[0126] Obtain historical CRBT videos and abnormal CRBT videos, and determine the pixel mean of the pixel points at the same position in each video frame in the historical CRBT video. For example, the historical CRBT video corresponds to two video frames, and the matrix corresponding to the pixel value of the pixel point in the first video frame is: The matrix corresponding to the pixel value of the pixel point of the second video frame is: Determine the pixel mean of each pixel in each video frame in the historical ringback tone video: Subtract the pixel value of each pixel point in the historical ringback tone video from the pixel point mean value of the pixel points at the same position to obtain a normal video; for example, the pixel value of the pixel point in the first video frame is subtracted from the pixel point mean value of the pixel points at the same position to obtain the matrix corresponding to the absolute value of the difference: The pixel value of the pixel point in the second video frame minus the pixel point mean corresponding to the pixel point at the same position, the matrix corresponding to the absolute value of the difference is obtained:

[0127] Similarly, the pixel mean of the pixels at the same position in each video frame in the abnormal ringback tone video is determined; the pixel value of the pixel of each video frame in the abnormal ringback tone video is subtracted from the pixel mean corresponding to the pixel at the same position to obtain the abnormal video.

[0128] The preset neural network model is trained based on the training set. Each video sequence input to the neural network model is N*227*227, where N is the total number of video frames of the video ringback tone and each video frame has 227*227 pixels. Each video sequence output by the neural network model is N*227*227.

[0129] like Figure 5 As shown, the neural network model includes a spatial autoencoder and a temporal autoencoder. The spatial autoencoder includes a spatial encoder and a spatial decoder, while the temporal autoencoder includes a temporal encoder and a temporal decoder. The spatial encoder receives a video sequence difference as input, processes N video frame differences, concatenates the N video frame differences encoded by the spatial encoder, and then inputs them into the temporal autoencoder for motion encoding. The temporal decoder and spatial decoder act as mirror images of the encoder.

[0130] The spatial encoder consists of two convolutional layers (Conv2D, two-dimensional convolution). In the first convolutional layer, the number of filters (i.e., convolution kernels) is set to 96, the filter shape is set to 7*7, the sliding step is set to 3 (the step size is the number of pixels that the filter passes through each time), the activation function is set to the linear rectifier function (Relu function), where Relu(x) = max(x, 0), and the padding is set to "same". That is, if the input data is not enough for the convolution kernel to scan, the input data will be padded with zeros. In the second convolutional layer, the number of filters is set to 64, the filter shape is set to 5*5, the sliding step is set to 2, the activation function is set to the linear rectifier function (Relu function), and the padding is set to "same".

[0131] The temporal autoencoder consists of a five-layer ConvLSTM2D (Convolutional LSTM) neural network. The filter shape is set to 3*3, and the number of filters is 64, 32, 16, 32, and 64, respectively, with the same padding.

[0132] The spatial decoder contains two deconvolution layers (Conv2DTanspose, two-dimensional deconvolution). In the first deconvolution layer: the number of filters is set to 128, the filter shape is set to 5*5, the sliding step is set to 2, the activation function is set to the linear rectification function, i.e., the ReLU function, and the padding is set to the same; in the second deconvolution layer: the number of filters is set to 128, the filter shape is set to 7*7, the sliding step is set to 3, the activation function is set to the linear rectification function, i.e., the ReLU function, and the padding is set to the same.

[0133] The optimizer uses an adaptive learning rate gradient descent algorithm to improve the learning speed of traditional gradient descent and imposes an adaptive learning rate constraint. The objective function uses a logarithmic loss function, as shown below: L(Y, P(Y|X)) = -logP(Y|X), where X represents the feature value of the training set samples and Y represents the true value of the training set samples. The number of training rounds is set to 1000, and the batch size is set to 64. The spatiotemporal coding neural network uses gradient descent to find the optimal weight values ​​that minimize the objective function. As the number of training rounds increases, the training error gradually decreases, and the model gradually converges. Finally, the weights of the neural network model are derived and used to determine the quality detection model. The training error of the trained neural network model can also be determined based on the test set. If the training error is less than a preset error value, the trained neural network model is used as the quality detection model.

[0134] In the technical solution of this embodiment, a training set and a test set are obtained. A preset neural network model is trained based on the training set, and the training error of the trained neural network model is determined based on the test set. When the training error is less than a preset threshold, the trained neural network model is used as a quality detection model. The quality detection model is obtained by training the training set, and the second pixel difference is determined using the quality detection model. This makes the calculation of the reconstruction error more accurate, thereby improving the accuracy of abnormal ringback tone video detection.

[0135] Reference Figure 6 , Figure 6 This is a fourth embodiment of the method for detecting the quality of a color ring back tone video according to the present invention, based on any one of the first to third embodiments, and before step S20, further comprising:

[0136] Step S40, obtaining the source ringback tone video and the screen recording ringback tone video;

[0137] Step S50, determining the pixel mean of the pixel points at the same position in each video frame in the source ring back tone video;

[0138] Step S60, subtracting the pixel value of each pixel in the source ring back tone video from the pixel value of the pixel at the same position to obtain the source video;

[0139] Step S70, determining the pixel mean of the pixel points at the same position in each video frame in the screen recording ringback tone video;

[0140] Step S80: Subtract the pixel value of each pixel in the screen recording ringback tone video from the pixel mean value corresponding to the pixel at the same position to obtain the screen recording video.

[0141] Specifically, before determining the first pixel difference between the source video and the screen recording video of the ring back tone video, the source ring back tone video is preprocessed to obtain the source video, and the screen recording ring back tone video is preprocessed to obtain the screen recording video. The source ring back tone video and the screen recording ring back tone video are obtained, and the pixel mean of each pixel point in each video frame in the source ring back tone video is determined. For example, the source ring back tone video corresponds to two video frames, and the matrix corresponding to the pixel value of the pixel point of the first video frame is: The matrix corresponding to the pixel value of the pixel point of the second video frame is: Determine the pixel mean of each pixel in each video frame of the source ringback tone video:

[0142] Subtract the pixel value of each pixel point in the source ringback tone video from the pixel value of the pixel point at the same position to obtain the source video; for example, the pixel value of the pixel point of the first video frame minus the pixel point mean value of the pixel point at the same position to obtain the matrix corresponding to the absolute value of the difference is The pixel value of the pixel point in the second video frame minus the pixel point mean corresponding to the pixel point at the same position, the matrix corresponding to the absolute value of the difference is obtained:

[0143] Similarly, determine the pixel mean of the pixel points at the same position of each video frame in the screen recording ringtone video; subtract the pixel mean corresponding to the pixel points at the same position from the pixel value of each video frame in the screen recording ringtone video to obtain the screen recording video.

[0144] In the technical solution of this embodiment, a source ringtone video and a screen recording ringtone video are obtained; the pixel mean of the pixels at the same position in each video frame of the source ringtone video is determined, and the pixel mean corresponding to the pixels at the same position is subtracted from the pixel value of each video frame in the source ringtone video to obtain the source video; the pixel mean of the pixels at the same position in each video frame of the screen recording ringtone video is determined, and the pixel mean corresponding to the pixels at the same position is subtracted from the pixel value of each video frame in the screen recording ringtone video to obtain the screen recording video. By preprocessing the video frames of the source ringtone video and the screen recording ringtone video to obtain the source video and the screen recording video, the reconstruction error determined based on the video frames of the source video and the screen recording video is made more accurate, thereby improving the accuracy of abnormal ringtone video detection.

[0145] Reference Figure 7 The present invention further provides a quality detection device for a ring back tone video, characterized in that the quality detection device for the ring back tone video comprises:

[0146] An acquisition module 100 is used to acquire a screen recording video corresponding to the played ringback tone video;

[0147] A determination module 200 is configured to determine a first pixel difference between a source video of a ring back tone video and the screen recording video;

[0148] The detection module 300 is configured to determine whether the CRBT video is an abnormal CRBT video according to the first pixel difference.

[0149] In one embodiment, in determining the first pixel difference between the source video of the ring back tone video and the screen recording video, the determining module 200 is specifically configured to:

[0150] Determine a first pixel difference between a video frame of a source video of the ring back tone video and a corresponding video frame of the screen recording video, where the first pixel difference includes pixel differences corresponding to each video frame.

[0151] In one embodiment, in determining whether the CRBT video is an abnormal CRBT video according to the first pixel difference, the detection module 300 is specifically configured to:

[0152] Inputting the first pixel difference into a preset quality detection model to obtain a second pixel difference output by the quality detection model, wherein the quality detection model is trained by normal video;

[0153] determining a reconstruction error according to the first pixel difference and the second pixel difference;

[0154] If the reconstruction error is greater than a preset threshold, determining that the CRBT video is an abnormal CRBT video;

[0155] If the reconstruction error is less than or equal to a preset threshold, it is determined that the CRBT video is a normal CRBT video.

[0156] In one embodiment, in determining the reconstruction error according to the first pixel difference and the second pixel difference, the detection module 300 is specifically configured to:

[0157] determining a Euclidean distance between the first pixel difference and the second pixel difference;

[0158] The reconstruction error is determined according to the Euclidean distance.

[0159] In one embodiment, before inputting the first pixel difference into a preset quality detection model, the detection module 300 is specifically configured to:

[0160] Obtaining a training set and a test set, wherein the training set includes normal videos, and the test set includes normal videos and abnormal videos;

[0161] Training a preset neural network model according to the training set, wherein the output of the neural network model is a second pixel difference of a normal video in the training set;

[0162] Determining a training error of the trained neural network model based on the test set;

[0163] When the training error is less than a preset error value, the trained neural network model is used as a quality detection model.

[0164] In one embodiment, before obtaining the training set and the test set, the detection module 300 is specifically configured to:

[0165] Obtain historical ringback tone videos and abnormal ringback tone videos;

[0166] Determine the pixel mean of the pixel points at the same position of each video frame in the historical ring back tone video;

[0167] Subtract the pixel value of each pixel point in the historical ringback tone video from the pixel point average value corresponding to the pixel point at the same position to obtain a normal video;

[0168] Determine the pixel mean of the pixel points at the same position of each video frame in the abnormal ring back tone video;

[0169] The abnormal video is obtained by subtracting the pixel value of each video frame in the abnormal ring back tone video from the pixel mean value corresponding to the pixel at the same position.

[0170] In one embodiment, in terms of obtaining the screen recording video corresponding to the played ringback tone video, the determining module 200 is specifically configured to:

[0171] Get the source ringback tone video and screen recording ringback tone video;

[0172] Determine the pixel mean of the pixel points at the same position of each video frame in the source ring back tone video;

[0173] Subtract the pixel value of each pixel in the source ring back tone video from the pixel mean value of the pixel at the same position to obtain the source video;

[0174] Determine the pixel mean of the pixel points at the same position of each video frame in the screen recording ringback tone video;

[0175] The pixel value of each video frame in the screen recording ringback tone video is subtracted from the pixel mean value corresponding to the pixel at the same position to obtain the screen recording video.

[0176] The present invention also provides a quality detection device for a color ring back tone video, which includes a memory, a processor, and a color ring back tone video quality detection program stored in the memory and executable on the processor. When the color ring back tone video quality detection program is executed by the processor, each step of the color ring back tone video quality detection method described in the above embodiment is implemented.

[0177] The present invention also provides a computer-readable storage medium storing a quality detection program for a ring back tone video. When the quality detection program for the ring back tone video is executed by a processor, each step of the method for quality detection of the ring back tone video described in the above embodiment is implemented.

[0178] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0179] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, system, article, or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, system, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, system, article, or device comprising the element.

[0180] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment system can be implemented by means of software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the existing technology, can be embodied in the form of a software product. The computer software product is stored in a computer-readable storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, parking management equipment, air conditioner, or network equipment, etc.) to execute the system described in each embodiment of the present invention.

[0181] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for detecting the quality of a ring back tone video, characterized in that: The quality detection method of the color ring back tone video includes: Get the screen recording video corresponding to the played ringback tone video; Determine a first pixel difference between a source video of a ringback tone video and a video frame corresponding to the screen recording video, wherein the first pixel difference includes a pixel difference corresponding to each video frame, wherein a first pixel value of each first pixel point in each video frame of the source video and a second pixel value of each second pixel point in each video frame of the screen recording video are determined, and the first pixel difference is determined based on a pixel difference between the first pixel value of each first pixel point of the source video and the second pixel value of the corresponding second pixel point of the screen recording video; Determine whether the CRBT video is an abnormal CRBT video according to the first pixel difference.

2. The method for detecting the quality of the ring back tone video according to claim 1, wherein: The step of determining whether the CRBT video is an abnormal CRBT video according to the first pixel difference includes: Inputting the first pixel difference into a preset quality detection model to obtain a second pixel difference output by the quality detection model, wherein the quality detection model is trained by normal video; determining a reconstruction error according to the first pixel difference and the second pixel difference; If the reconstruction error is greater than a preset threshold, determining that the CRBT video is an abnormal CRBT video; If the reconstruction error is less than or equal to a preset threshold, it is determined that the CRBT video is a normal CRBT video.

3. The method for detecting the quality of the ring back tone video according to claim 2, wherein: The step of determining the reconstruction error according to the first pixel difference and the second pixel difference comprises: determining a Euclidean distance between the first pixel difference and the second pixel difference; The reconstruction error is determined according to the Euclidean distance.

4. The method for detecting the quality of the color ring back tone video according to claim 2, wherein: Before the step of inputting the first pixel difference into a preset quality detection model, the method further includes: Obtaining a training set and a test set, wherein the training set includes normal videos, and the test set includes normal videos and abnormal videos; Training a preset neural network model according to the training set, wherein the output of the neural network model is a second pixel difference of a normal video in the training set; Determining a training error of the trained neural network model based on the test set; When the training error is less than a preset error value, the trained neural network model is used as a quality detection model.

5. The method for detecting the quality of the color ring back tone video according to claim 4, wherein: Before the step of obtaining the training set and the test set, the method further includes: Obtain historical ringback tone videos and abnormal ringback tone videos; Determine the pixel mean of the pixel points at the same position of each video frame in the historical ring back tone video; Subtract the pixel value of each pixel point in the historical ringback tone video from the pixel point average value corresponding to the pixel point at the same position to obtain a normal video; Determine the pixel mean of the pixel points at the same position of each video frame in the abnormal ring back tone video; The abnormal video is obtained by subtracting the pixel value of each video frame in the abnormal ring back tone video from the pixel mean value corresponding to the pixel at the same position.

6. The method for detecting the quality of the color ring back tone video according to claim 1, wherein: The step of obtaining the screen recording video corresponding to the played ringback tone video includes: Get the source ringback tone video and screen recording ringback tone video; Determine the pixel mean of the pixel points at the same position of each video frame in the source ring back tone video; Subtract the pixel value of each pixel in the source ring back tone video from the pixel mean value of the pixel at the same position to obtain the source video; Determine the pixel mean of the pixel points at the same position of each video frame in the screen recording ringback tone video; The pixel value of each video frame in the screen recording ringback tone video is subtracted from the pixel mean value corresponding to the pixel at the same position to obtain the screen recording video.

7. A quality detection device for a ring back tone video, characterized in that: The quality detection device of the color ring back tone video includes: The acquisition module is used to obtain the screen recording video corresponding to the played ringback tone video; a determination module, configured to determine a first pixel difference between a source video of a ringback tone video and a video frame corresponding to the screen recording video, wherein the first pixel difference includes a pixel difference corresponding to each video frame, wherein a first pixel value of each first pixel point in each video frame of the source video and a second pixel value of each second pixel point in each video frame of the screen recording video are determined, and the first pixel difference is determined based on the pixel difference between the first pixel value of each first pixel point of the source video and the second pixel value of the corresponding second pixel point of the screen recording video; A detection module is configured to determine whether the CRBT video is an abnormal CRBT video according to the first pixel difference.

8. A quality detection device for a ring back tone video, characterized in that: The CRBT video quality detection device includes a memory, a processor, and a CRBT video quality detection program stored in the memory and executable on the processor. When the CRBT video quality detection program is executed by the processor, each step of the CRBT video quality detection method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a quality detection program for a CRBT video. When the CRBT video quality detection program is executed by a processor, each step of the quality detection method for a CRBT video according to any one of claims 1 to 6 is implemented.

Images