Malfunction video processing method and device, electronic equipment and storage medium

Abstract

The application provides a processing method and device for a down recording video, electronic equipment and a storage medium. The method comprises the following steps: determining a text area of a down recording video, performing color clustering on an image in the text area to obtain a plurality of single-color layer images; obtaining a text connected component of each layer image, performing text recognition on the text connected component to obtain a recognition result of each layer image; converting the recognition text corresponding to the recognition result into a storage document, and saving the storage document in a log system of a server. The application can store the recognition result of the down recording video, and the recognition result can be queried to determine the cause of the down of the server, so that the down problem can be located and excluded, and the cause query efficiency of the down of the server is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of servers, and in particular to a processing method and device for a server crash video, an electronic device and a storage medium. BACKGROUND

[0002] With the continuous growth of netizens, a stable network environment is increasingly important. Currently, the positioning of server crash problems in the computer room still mainly relies on text logs or simple image information, and it is extremely inconvenient to exclude the positioning of the crash problem.

[0003] Currently, existing counting analyzes the black box log recorded by the BMC to analyze the crash reason, but this way has less information and simple information dimensions. Or through the third-party graphics card to record image information, but through the third-party graphics card to record images, it is not convenient to implement support for old devices.

[0004] Therefore, how to position and exclude the server crash problem is a technical problem that needs to be solved at present. SUMMARY

[0005] The present application provides a processing method and device for a server crash video, an electronic device and a storage medium, to solve the above-mentioned defects in the prior art, to store the recognition result, to facilitate query to determine the reason for the server crash, to position and exclude the server crash problem, and to improve the query efficiency of the server crash reason.

[0006] The present application provides a processing method for a server crash video, comprising:

[0007] determining a text area of the server crash video, performing color clustering on the image in the text area to obtain a plurality of single-color layer images;

[0008] obtaining a text connected component of each of the layer images, performing text recognition on the text connected component to obtain a recognition result of each of the layer images;

[0009] converting the recognition text corresponding to the recognition result into a storage document, and saving the storage document in a log system of a server.

[0010] According to the processing method for a server crash video provided by the present application, the recognition text corresponding to the recognition result is converted into a storage document, comprising:

[0011] obtaining the generation time of the recognition text corresponding to each of the recognition results;

[0012] sorting each of the generation times in reverse order of time to determine a time sequence;

[0013] Combine the identified text with the time sequence to convert the storage document.

[0014] According to the application, the text connected components of each layer image are obtained, and the method comprises the following steps:

[0015] Based on the connected component analysis method, the candidate text connected components of each layer image are obtained.

[0016] Based on the geometric constraint of the text area, the candidate text connected components are screened, and the non-text connected components in the candidate text connected components are removed to obtain the text connected components.

[0017] According to the application, the processing method of the down machine video further comprises the following steps before determining the text area of the down machine video:

[0018] Real-time record the operation request received by the server, record the operation request in time sequence to obtain record information;

[0019] Based on the time span associated with the record information, the set time interval and the down machine trigger information, the record information is updated to obtain target record information;

[0020] The storage document is saved in the log system of the server, and the log system of the server comprises:

[0021] Based on the storage document and the target record information, a query log is generated; the query log is used to determine the time when the server appears down and determine the reason why the server appears down;

[0022] The query log is stored in the log system.

[0023] According to the application, the processing method of the down machine video further comprises the following steps before determining the text area of the down machine video:

[0024] When the time span associated with the record information exceeds the set time interval and the down machine trigger information is not identified in the time interval, the record information is deleted;

[0025] When the time span associated with the record information is within the set time interval and the down machine trigger information is identified in the time interval, the record information is saved as target record information; wherein the target record information comprises record data matched with the down machine video, and the record data is used to integrate the identified text obtained based on the down machine video and the image of the down machine video in time sequence.

[0026] According to the processing method of the off-line video provided by the application, after the text connected components of the text recognition result of each layer image are obtained, the method further comprises:

[0027] Obtaining the layer relationship between each layer image;

[0028] Verifying each recognition result based on the layer relationship.

[0029] According to the processing method of the off-line video provided by the application, the method for determining the text area of the off-line video comprises:

[0030] Obtaining the key frame of the off-line video, and performing text positioning on the image corresponding to the key frame to determine a target area;

[0031] Filtering the target area to determine a background area and the text area in the target area.

[0032] The application further provides a processing device of off-line video, comprising:

[0033] A clustering module for determining the text area of the off-line video, and performing color clustering on the image in the text area to obtain a plurality of single-color layer images;

[0034] An identification module for obtaining the text connected components of each layer image, and performing text recognition on the text connected components to obtain the recognition result of each layer image;

[0035] A storage module for converting the recognition text corresponding to the recognition result into a storage document, and saving the storage document in the log system of a server.

[0036] The application further provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the processing method of the off-line video as described above.

[0037] The application further provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the processing method of the off-line video as described above.

[0038] The application further provides a computer program product comprising a computer program, wherein the computer program is executable by a processor to implement the processing method of the off-line video as described above.

[0039] The application provides a processing method and device of a down recording video, electronic equipment and a storage medium. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0041] Figure 1 Fig. 1 is a flow diagram of the processing method of the down recording video provided by the application;

[0042] Figure 2 Fig. 2 is a structural diagram of the processing device of the down recording video provided by the application;

[0043] Figure 3 Fig. 3 is a structural diagram of the electronic equipment provided by the application. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the application more clear, the technical solutions in the application will be described clearly and completely in the following with reference to the drawings in the application. Obviously, the described embodiments are some embodiments of the application, not all the embodiments. Based on the embodiments in the application, all the other embodiments obtained by those skilled in the art without any creative effort belong to the protection scope of the application.

[0045] It should be first noted that the execution subject of the processing method of the off-line video provided by the present application can be an electronic device, a component in the electronic device, an integrated circuit, or a chip. The electronic device can be a mobile electronic device or a non-mobile electronic device. Exemplarily, the mobile electronic device can be a mobile phone, a tablet computer, a notebook computer, a palm computer, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), and the non-mobile electronic device can be a server, a network attached storage (NAS), or a personal computer (PC), and the present application is not limited in this regard.

[0046] With reference to Figure 1 The processing method of the off-line video provided by the present application comprises the following steps:

[0047] In step 110, the text region of the off-line video is determined, the images in the text region are color clustered, and a plurality of single-color layer images are obtained.

[0048] In step 120, the text connected components of each layer image are obtained, the text recognition is performed on the text connected components, and the recognition results of each layer image are obtained.

[0049] In step 130, the recognized text corresponding to the recognition results is converted into a storage document, and the storage document is saved in the log system of a server.

[0050] In the above step 110, the key frame extraction can be performed on each shot boundary of the off-line video, the off-line video is analyzed frame by frame, a plurality of key frames of the off-line video are obtained according to the mutation information of the video frames, and the images corresponding to each key frame are obtained.

[0051] Then, the text region positioning is performed on the images corresponding to each key frame, a text-containing region is obtained, and the regions can be further screened, such as filtering processing, the regions can be further classified, and the more accurate text region and background region are determined.

[0052] The image is processed by using the color clustering method for the text area to obtain an image of multiple single-color layers. It should be noted that the color clustering algorithm is usually based on three dimensions of R (Red), G (Green), and B (Blue) to calculate and perform clustering and division in the RGB color space. In other implementation processes, the color clustering algorithm can also be used for image denoising, image segmentation, or image color layering, etc.

[0053] For example, assuming that the RGB of two colors are (R1, G1, B1) and (R2, G2, B2), if the value calculated by (R1-R2+1)*(G1-G2+1)*(B1-B2+1) is greater than a pre-set value, it is considered that the two colors are different hues.

[0054] In step 120, the text connected components of each layer image are obtained by using the connected component analysis method. The connected component analysis, also known as connected component labeling, spot extraction, or region labeling, is an algorithm application of graph theory, which is used to determine the connectivity of "spot" areas in a binary image.

[0055] Then, the obtained text connected components are subjected to character recognition. In this embodiment, the character recognition method can be OCR recognition. OCR (optical character recognition) character recognition refers to a process in which an electronic device (such as a scanner or a digital camera) checks characters printed on paper and then translates the shapes into computer characters by using character recognition methods; that is, a process of scanning text materials and then analyzing and processing image files to obtain character and layout information. In this embodiment, the specific recognition process can use an OCR character package to recognize the text connected components, or a trained neural network model can be used to recognize the text connected components, so as to obtain the character recognition result of each layer image.

[0056] In step 130, the character recognition result of each layer image obtained above is converted into a text form to obtain a storage document, and the storage document is saved to the log system of the server BMC for user query. In some optional embodiments, the storage document can include system time, IP address, user operation information, etc.

[0057] When the server is in a downtime state, the user can directly query the storage document in the log system, and by reading the storage document, the text information and time information in the downtime video can be determined, and the reason for the downtime of the server can be determined according to the read content.

[0058] The application provides a processing method for a down recording video, which comprises the following steps: determining a text area of the down recording video, performing color clustering on images in the text area to obtain a plurality of single-color layer images, obtaining text connected components of each layer image, performing text recognition on the text connected components to obtain recognition results of each layer image, and finally converting the recognition text corresponding to the recognition results into a storage document and saving the storage document in a log system of a server.

[0059] In some optional embodiments, the converting the recognition text corresponding to the recognition results into a storage document comprises:

[0060] obtaining generation times of the recognition text corresponding to each of the recognition results;

[0061] sequencing the generation times in a reverse order of time to determine a time sequence;

[0062] converting the recognition text and the time sequence into the storage document.

[0063] Specifically, in order to facilitate users to locate and eliminate a server down more conveniently and in more detail, the embodiment provides a process of determining a storage document in combination with a generation time.

[0064] In the process of converting the recognition results of the text into the recognition text, the generation times of the recognition text corresponding to each of the recognition results can be obtained, which can be denoted as T1, T2, T3 and T4. The generation times T1, T2, T3 and T4 are sequenced in a reverse order of time, i.e., in a reverse order according to the sequence of the generation times, to determine a time sequence. Finally, the recognition text is converted into the storage document in combination with the time sequence.

[0065] The processing method for the down recording video provided by the embodiment of the application saves the storage document with the time sequence into a log system. When the server is down, a user can directly query the storage document in the log system and determine the text information and the time information in the down video by reading the storage document, so as to directly determine the time when the server is down according to the read content, thereby facilitating timely service area maintenance.

[0066] In some optional embodiments, the obtaining the text connected components of each of the layer images comprises:

[0067] obtaining candidate text connected components of each of the layer images based on a connected component analysis method;

[0068] Screen the candidate text connected components based on the geometric constraint of the character region, eliminate the non-text connected components in the candidate text connected components, and obtain the text connected components.

[0069] It can be understood that the embodiment is to ensure the accuracy of the text connected components, and the process of screening the candidate text.

[0070] Firstly, the candidate text connected components of each layer image are obtained by using the connected component analysis method. It should be noted that the candidate text connected components include text connected components and non-text connected components. Then, the candidate text connected components are screened based on the geometric constraint of the character region, the non-text connected components in the candidate text connected components are eliminated, and the text connected components are retained.

[0071] It should be noted that the geometric constraint condition in the embodiment is pre-set. The geometric constraint is to constrain geometric elements through geometric relations. For example, if the candidate text connected components in the embodiment are less than a set first threshold in horizontal or vertical angle, and the geometric distance is less than a set second threshold, the candidate text connected components can be regarded as text connected components, otherwise, the candidate text connected components are regarded as non-text connected components.

[0072] The processing method of the down machine video provided by the embodiment of the application screens the candidate text connected components through the geometric constraint, so as to further accurately determine the text connected components, avoid the process of text recognition on the non-text connected components, and further improve the overall text recognition efficiency.

[0073] In some optional embodiments, before the determining of the character region of the down machine video, the method further includes:

[0074] Real-time record the operation request received by the server, record the operation request in time sequence to obtain record information;

[0075] Based on the time span associated with the record information, the set time interval and the down machine triggering information, the record information is updated to obtain target record information.

[0076] It can be understood that the embodiment is a step performed before the server is down. Firstly, the operation request received by the server needs to be recorded in real time. The operation request can refer to the control instruction, interaction data and the like sent by the user end to the server. Then, the operation request is recorded in time sequence to obtain the record information of the operation request.

[0077] The recording information is updated according to a preset target time interval, and in actual application, the recording information can be updated once per hour. In the case of server downtime, the current recording information is saved as target recording information. It can be understood that when the recording information is updated, the recording information one hour ago is deleted, and the new recording information is saved as target recording information.

[0078] Further, the log system for saving the storage document on the server comprises:

[0079] The query log is generated based on the storage document and the target recording information, and is used to determine the time of server downtime and the cause of server downtime;

[0080] The query log is stored in the log system.

[0081] It can be understood that the embodiment generates the query log by using the updated target recording information and the storage document with time sequence. When the server is down, the user can directly determine the time of server downtime through the query log, determine the corresponding operation information based on the time and the target recording information, and thus determine the cause of server downtime.

[0082] The processing method for server downtime video provided by the embodiment ensures the real-time performance of the recording information by recording the operation information of the user end before server downtime and updating the recording information every certain period of time, and then saves the recording information and the storage document with time sequence of the recognized text as a query log, so that the cause and time of server downtime can be directly determined through the query log when the server is down, and the efficiency of positioning and excluding the server downtime problem is further improved.

[0083] In some optional embodiments, the recording information is updated based on the time span associated with the recording information, the set time interval, and the downtime trigger information, to obtain target recording information, comprising:

[0084] When the time span associated with the recording information exceeds the set time interval and no downtime trigger information is recognized in the time interval, the recording information is deleted;

[0085] When the time span associated with the recording information is within the set time interval and the downtime trigger information is recognized in the time interval, the recording information is saved as target recording information; wherein the target recording information comprises recording data matched with the server downtime video, and the recording data is used to integrate the recognized text obtained based on the server downtime video and the image of the server downtime video in time sequence.

[0086] That is, when the time span associated with the record information exceeds a set time interval, and within one hour of the time interval, no downtime trigger information appears, that is, no server downtime phenomenon appears, the record information is deleted.

[0087] When the time span associated with the record information is within a set time interval, and within one hour of the time interval, downtime trigger information appears, that is, a server downtime phenomenon appears, the record information within the one hour needs to be saved as target record information, so as to facilitate subsequent conversion of the target record information into a log file for storage.

[0088] The target record information includes record data matched with the downtime video recording video, and the record data is used to integrate the recognized text based on the downtime video recording video and the images of the downtime video recording video in chronological order.

[0089] In some optional embodiments, after the text recognition on the text connected components is performed to obtain the recognition results of the layer images, the method further includes:

[0090] Obtaining a layer relationship between the layer images;

[0091] Verifying each of the recognition results based on the layer relationship.

[0092] In this embodiment, the layer relationship between each of the layer images is obtained, and then each of the recognition results is verified based on the layer relationship, that is, each of the layer images is verified with each other, so that the accuracy of the recognition results is improved.

[0093] In some optional embodiments, the determination of the text region of the downtime video recording video includes:

[0094] Obtaining a key frame of the downtime video recording video, performing text positioning on an image corresponding to the key frame to determine a target region;

[0095] Filtering the target region to determine a background region and the text region in the target region.

[0096] Specifically, in this embodiment, first, key frame extraction is performed on each shot boundary of the downtime video recording video. According to frame-by-frame analysis of the downtime video recording video, a plurality of key frames of the downtime video recording video are obtained according to the mutation information of the video frames, so as to obtain an image corresponding to each key frame.

[0097] Then, text region positioning is performed on the image corresponding to each key frame to obtain a target region containing text. The target region is further filtered and classified, so as to determine a more accurate text region and a background region.

[0098] It should be noted that the filtering process in the embodiment can be filtering the text region by a Gabor filter. The Gabor filter can extract texture information in different directions. The Gabor filter is not sensitive to light changes, can provide good adaptability to light changes, can tolerate a certain degree of image rotation and deformation, and has a certain robustness to light and posture

[0099] The processing method of the crash video provided by the application determines the image corresponding to the key frame by key frame extraction of the crash video, then determines the approximate text region, and accurately determines the text region and the background region through the filtering process, so that the text recognition can be performed on the text region, the text recognition result can be determined more efficiently and stored, and the query efficiency of the server crash reason is further improved.

[0100] The processing device of the crash video provided by the application is described below, and the processing device of the crash video described below can be correspondingly referred to the processing method of the crash video described above.

[0101] Reference Figure 2 The processing device of the crash video provided by the application includes the following modules:

[0102] The clustering module 210 is configured to determine the text region of the crash video, and perform color clustering on the image in the text region to obtain a plurality of single-color layer images.

[0103] The recognition module 220 is configured to obtain the text connected component of each layer image, and perform text recognition on the text connected component to obtain the recognition result of each layer image.

[0104] The storage module 230 is configured to convert the recognized text corresponding to the recognition result into a storage document, and save the storage document in the log system of the server.

[0105] In the above clustering module 210, the boundary of each shot of the crash video needs to be detected first. The detection process can be key frame extraction. The crash video can be analyzed frame by frame, and a plurality of key frames of the crash video can be obtained according to the mutation information of the video frame, so as to obtain the image corresponding to each key frame.

[0106] Then, the image corresponding to each key frame is positioned to obtain a region containing text, and these regions can be further screened, such as filtering, and these regions can be further classified to determine the more accurate text region and the background region.

[0107] The image is processed by using the color clustering method for the text area to obtain an image of multiple single-color layers. It should be noted that the color clustering algorithm is usually based on three dimensions of R (Red), G (Green), and B (Blue) to calculate and perform clustering division in the RGB color space. In other implementation processes, the color clustering algorithm can also be used for image denoising, image segmentation, or image color layering, etc.

[0108] For example, assuming that the RGB of two colors are (R1, G1, B1) and (R2, G2, B2), if the value calculated by (R1-R2+1)*(G1-G2+1)*(B1-B2+1) is greater than a pre-set value, it is considered that the two colors are different hues.

[0109] In the above-mentioned identification module 220, the text connected components of each layer image are obtained by using the connected component analysis method. The connected component analysis is also called connected component labeling, spot extraction, or region labeling, which is an algorithm application of graph theory and is used to determine the connectivity of “spot”-shaped regions in a binary image.

[0110] Then, the obtained text connected components are subjected to character recognition. In this embodiment, the character recognition mode can be OCR recognition. OCR (optical character recognition) character recognition refers to a process in which an electronic device (such as a scanner or a digital camera) checks characters printed on paper and then translates the shapes into computer characters by using character recognition methods; that is, a process of scanning text materials and then analyzing and processing image files to obtain character and layout information. In this embodiment, the specific recognition process can use an OCR character package to recognize the text connected components, or a trained neural network model can be used to recognize the text connected components, so as to obtain the character recognition result of each layer image.

[0111] In the above-mentioned storage module 230, the character recognition result of each layer image obtained above is converted into a text form to obtain a storage document, and the storage document is saved to the log system of the server BMC for user query. In some optional embodiments, the storage document can include system time, IP address, user operation information, etc.

[0112] When the server is in a downtime situation, the user can directly query the storage document in the log system, and by reading the storage document, the character information and time information in the downtime video can be determined, and the cause of the server downtime can be determined according to the read content.

[0113] The application provides a processing device for a crash video, which determines a text area of a crash video, performs color clustering on images in the text area, obtains a plurality of single-color layer images, obtains text connected components of each layer image, performs text recognition on the text connected components, obtains recognition results of each layer image, converts the recognition results into storage documents, and saves the storage documents in a log system of a server.

[0114] In some optional embodiments, the storage module is specifically configured to:

[0115] obtain generation times of the recognition texts corresponding to the recognition results;

[0116] sort the generation times in a time reverse order to determine a time sequence;

[0117] convert the recognition texts and the time sequence into the storage documents.

[0118] Specifically, in order to facilitate a user to locate and exclude a server crash in a more detailed and convenient manner, the embodiment provides a process of determining a storage document in combination with a generation time.

[0119] In the process of converting the recognition results of the texts into the recognition texts, the generation times of the recognition texts corresponding to each recognition result can be obtained, which can be denoted as T1, T2, T3, T4, etc. The generation times T1, T2, T3, T4 are sorted in a time reverse order according to the order of the generation times to determine a time sequence. Finally, the recognition texts are converted into the storage documents in combination with the time sequence.

[0120] The processing device for the crash video provided in the embodiment of the application saves the storage documents with the time sequence into a log system. When a server crashes, a user can directly query the storage documents in the log system, determines text information and time information in the crash video by reading the storage documents, and directly determines the time when the server crashes according to the read content, so as to timely maintain a service area.

[0121] In some optional embodiments, the recognition module is specifically configured to:

[0122] obtain candidate text connected components of each layer image based on a connected component analysis method;

[0123] Screen the candidate text connected components based on the geometric constraint of the character region, eliminate the non-text connected components in the candidate text connected components, and obtain the text connected components.

[0124] It can be understood that the embodiment is to ensure the accuracy of the text connected components, and the process of screening the candidate text.

[0125] Firstly, the candidate text connected components of each layer image are obtained by using the connected component analysis method. It should be noted that the candidate text connected components include text connected components and non-text connected components. Then, the candidate text connected components are screened based on the geometric constraint of the character region, the non-text connected components in the candidate text connected components are eliminated, and the text connected components are retained.

[0126] It should be noted that the geometric constraint condition in the embodiment is pre-set. The geometric constraint is to constrain geometric elements through geometric relations. For example, if the candidate text connected components in the embodiment are less than the set first threshold value in the horizontal or vertical angle, and the geometric distance is less than the set second threshold value, the candidate text connected components can be regarded as the text connected components, otherwise, the candidate text connected components are regarded as the non-text connected components.

[0127] The processing device of the video recording provided by the embodiment of the application screens the candidate text connected components through the geometric constraint, so as to further accurately determine the text connected components, avoid the process of character recognition on the non-text connected components, and further improve the overall character recognition efficiency.

[0128] In some optional embodiments, an updating module is further included, and the updating module is configured to:

[0129] Record the operation request received by the server in real time, record the operation request in time sequence to obtain record information;

[0130] Update the record information based on the time span associated with the record information, the set time interval, and the downtime triggering information to obtain target record information.

[0131] It can be understood that the embodiment is a step performed before the server appears downtime. Firstly, the operation request received by the server needs to be recorded in real time, and the operation request can refer to the control instruction, interaction data and the like sent by the user end to the server. Then, the operation request is recorded in time sequence to obtain the record information of the operation request.

[0132] The recording information is updated according to a preset target time interval, and in actual application, the recording information can be updated once per hour. In the case of server downtime, the current recording information is saved as target recording information. It can be understood that when the recording information is updated, the recording information one hour ago is deleted, and the new recording information is saved as target recording information.

[0133] In some optional embodiments, the storage module is further used for:

[0134] generating a query log based on the storage document and the target recording information; the query log is used to determine the time when the server is down and determine the reason why the server is down;

[0135] storing the query log in the log system.

[0136] It can be understood that the embodiment generates a query log by using the updated target recording information and the storage document with time sequence. When the server is down, the user can directly determine the time when the server is down through the query log, determine the corresponding operation information based on the time and the target recording information, and thus determine the reason why the server is down.

[0137] The processing device for server downtime video provided by the embodiment ensures the real-time performance of the recording information by recording the operation information of the user terminal before the server is down and updating the recording information every certain period of time, and then saves the recording information and the storage document with time sequence of the recognized text as a query log, so that the reason and time of server downtime can be directly determined through the query log when the server is down, and the efficiency of positioning and excluding the server downtime problem is further improved.

[0138] In some optional embodiments, the update module is further used for:

[0139] when the time span associated with the recording information exceeds the set time interval and no downtime trigger information is recognized in the time interval, deleting the recording information;

[0140] when the time span associated with the recording information is within the set time interval and downtime trigger information is recognized in the time interval, saving the recording information as target recording information; wherein the target recording information includes recording data matched with the server downtime video, and the recording data is used to integrate the recognized text obtained based on the server downtime video and the image of the server downtime video in time sequence.

[0141] That is, when the time span associated with the record information exceeds a set time interval, and within one hour of the time interval, no downtime trigger information appears, that is, no server downtime phenomenon appears, the record information is deleted.

[0142] When the time span associated with the record information is within a set time interval, and within one hour of the time interval, downtime trigger information appears, that is, a server downtime phenomenon appears, the record information within the one hour needs to be saved as target record information, so as to facilitate subsequent conversion of the target record information into a log file for storage.

[0143] The target record information includes record data matched with the downtime video recording video, and the record data is used to integrate the recognized text obtained based on the downtime video recording video and the image of the downtime video recording video in time sequence.

[0144] In some optional embodiments, the verification module is further configured to:

[0145] Obtain a layer relationship between each of the layer images;

[0146] Verify each of the recognition results based on the layer relationship.

[0147] In this embodiment, the layer relationship between each of the layer images is obtained, and then each of the recognition results is verified based on the layer relationship, that is, each of the layer images is verified with each other, so as to improve the accuracy of the recognition results.

[0148] In some optional embodiments, the clustering module is specifically configured to:

[0149] Obtain a key frame of the downtime video recording video, perform text positioning on an image corresponding to the key frame, and determine a target region;

[0150] Filter the target region, and determine a background region and the text region in the target region.

[0151] Specifically, in this embodiment, first, key frame extraction is performed on each shot boundary of the downtime video recording video, frame-by-frame analysis is performed on the downtime video recording video, and a plurality of key frames of the downtime video recording video are obtained according to the mutation information of the video frames, so as to obtain an image corresponding to each key frame.

[0152] Then, text region positioning is performed on the image corresponding to each key frame, to obtain a target region containing text, and the target region is further filtered and classified, so as to determine a more accurate text region and a background region.

[0153] It should be noted that the filtering process in the embodiment can be optional filtering of the character region by a Gabor filter. The Gabor filter can extract texture information in different directions. The Gabor filter is not sensitive to light changes, can provide good adaptability to light changes, can tolerate a certain degree of image rotation and deformation, and has a certain robustness to light and posture

[0154] The processing device of the crash video provided by the application determines the image corresponding to the key frame by key frame extraction of the crash video, then determines the approximate character region, and accurately determines the character region and the background region through the filtering process, so that the character recognition can be performed on the character region, the character recognition result can be determined more efficiently and stored, and the query efficiency of the server crash reason is further improved.

[0155] Figure 3 An example of an entity structure diagram of an electronic device is shown in Figure 3 As shown, the electronic device can include a processor 310, a communication interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communication interface 320 and the memory 330 communicate with each other through the communication bus 340. The processor 310 can call the logical instructions in the memory 330 to execute the processing method of the crash video, which includes:

[0156] Determine the character region of the crash video, color cluster the image in the character region, and obtain a plurality of single-color layer images;

[0157] Obtain the text connected component of each layer image, perform character recognition on the text connected component, and obtain the recognition result of each layer image;

[0158] Convert the recognized character corresponding to the recognition result into a storage document, and save the storage document in the log system of the server.

[0159] In addition, the logic instructions in the memory 330 described above can be implemented in the form of a software function unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the part of the prior art that contributes essentially or the part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0160] In another aspect, the present application also provides a computer program product, which comprises a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program can be executed by a processor to enable a computer to execute the processing method of the off-line video provided by the above-mentioned methods, which comprises:

[0161] determining a text region of the off-line video, performing color clustering on images in the text region to obtain a plurality of single-color layer images;

[0162] obtaining text connected components of each of the layer images, performing text recognition on the text connected components to obtain recognition results of each of the layer images;

[0163] converting the recognized text corresponding to the recognition results into a storage document, and saving the storage document in a log system of a server.

[0164] In another aspect, the present application also provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the processing method of the off-line video provided by the above-mentioned methods, which comprises:

[0165] determining a text region of the off-line video, performing color clustering on images in the text region to obtain a plurality of single-color layer images;

[0166] obtaining text connected components of each of the layer images, performing text recognition on the text connected components to obtain recognition results of each of the layer images;

[0167] converting the recognized text corresponding to the recognition results into a storage document, and saving the storage document in a log system of a server.

[0168] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0169] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and necessary universal hardware platforms, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of software products, and the computer software products can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and include a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0170] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for processing a crash recording, characterized in that The method comprises: determining a text region of a down-time video, performing color clustering on an image in the text region to obtain a plurality of single-color layer images; obtaining text connected components of each of the layer images, and performing text recognition on the text connected components to obtain recognition results of each of the layer images; converting recognition text corresponding to the recognition results into a storage document, and saving the storage document in a log system of a server; wherein the obtaining of the text connected components of each of the layer images comprises: obtaining candidate text connected components of each of the layer images based on a connected component analysis method; screening the candidate text connected components based on geometric constraints of the text region, and removing non-text connected components in the candidate text connected components to obtain the text connected components; wherein, when the candidate text connected components are less than a set first threshold in horizontal or vertical angle, and the geometric distance of the candidate text connected components is less than a set second threshold, the candidate text connected components are taken as the text connected components; wherein, after the text recognition on the text connected components to obtain the recognition results of each of the layer images, the method further comprises: obtaining layer relationships between each of the layer images; verifying each of the recognition results based on the layer relationships.

2. The method of claim 1, wherein, The conversion of the recognition text corresponding to the recognition results into the storage document comprises: obtaining generation times of the recognition text corresponding to each of the recognition results; sorting each of the generation times in time reverse order to determine a time sequence; converting the recognition text in combination with the time sequence into the storage document.

3. The method of claim 1, wherein, Before the determination of the text region of the down-time video, the method further comprises: recording operation requests received by the server in real time, recording the operation requests in time sequence to obtain record information; updating the record information based on a time span associated with the record information, a set time interval, and down-time triggering information to obtain target record information; The saving of the storage document in the log system of the server comprises: generating a query log based on the storage document and the target record information; the query log is used to determine a time when the server appears to be down and determine a reason why the server appears to be down; storing the query log in the log system.

4. The processing method of claim 3, wherein, The updating of the record information based on the time span associated with the record information, the set time interval, and the down-time triggering information to obtain the target record information comprises: deleting the record information when the time span associated with the record information exceeds the set time interval and no down-time triggering information is identified in the time interval; saving the record information as the target record information when the time span associated with the record information is within the set time interval and down-time triggering information is identified in the time interval; wherein, the target record information comprises record data matched with the down-time video, and the record data is used to integrate the recognition text obtained based on the down-time video and the image of the down-time video in time sequence.

5. The method of claim 1, wherein, The determination of the text region of the down-time video comprises: Acquire key frames of the down video, locate texts in images corresponding to the key frames, and determine a target region; Filter the target region, and determine a background region and the text region in the target region.

6. A processing device for a crash recording, characterized in that The method comprises the following steps: A clustering module is configured to determine a text region of a down video, perform color clustering on images in the text region, and obtain a plurality of single-color layer images; An identification module is configured to acquire text connected components of each of the layer images, perform text recognition on the text connected components, and obtain identification results of each of the layer images; A storage module is configured to convert identification texts corresponding to the identification results into a storage document, and save the storage document in a log system of a server; The identification module is specifically configured to: Acquire candidate text connected components of each of the layer images based on a connected component analysis method; Filter the candidate text connected components based on geometric constraints of the text region, eliminate non-text connected components in the candidate text connected components, and obtain the text connected components; wherein, when the candidate text connected components are less than a first threshold value in horizontal or vertical angles, and geometric distances of the candidate text connected components are less than a second threshold value, the candidate text connected components are taken as the text connected components; The down video processing device further comprises a verification module, and the verification module is configured to: Acquire layer relationships between each of the layer images; Verify each of the identification results based on the layer relationships.

7. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the down video processing method according to any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the down video processing method according to any one of claims 1 to 5.

Images