KR101568680B1 - Data restoration method using data fragment classification - Google Patents

Abstract

The present invention relates to a data restoration method using data fragment classification. More preferably, the data restoration method includes: a step of an input unit for receiving input of restoration target data which are damaged and need to be restored; a step of a fragment classification unit for classifying fragments, related to the restoration target data, among data fragments pre-stored in a recording medium; a step of a fragment reassembling unit for reassembling fragments, which are irregularly fragmented, among the classified fragments; and a step of an image frame extraction unit for extracting an image frame with respect to the restoration target data based on the result of reassembling the fragments. According to the data restoration method using data fragments, data can be restored in an easy manner even when data fragments to be restored are irregularly fragmented as an image frame is extracted by classifying data fragments, related to data fragments to be restored, and reassembling the classified data fragments.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data restoration method,

The present invention relates to a data restoration method using data fragment classification, and more particularly, to a data restoration method using data fragment classification that can easily restore deleted or damaged data in a storage medium.

As IT technology develops rapidly, IT technology is converged and widely used in various fields. In particular, as many people access the wired and wireless Internet anytime and anywhere and use digital devices, various technologies for storing more multimedia data in a storage medium are required.

In particular, in order to effectively use a storage medium, a file system for managing data such as storing, deleting, and modifying data is required.

1 is a diagram illustrating a method of recording data in a storage medium in a digital device using a file system according to the related art.

As shown in FIG. 1 (a), additional information such as name, size, and location of data or a file is stored in a metadata area used by the file system, and actual data is stored in a storage area.

In this way, a file system uses a unit of a certain size to write data to a storage medium, which is called a data fragment. For example, if one file (File 1) consists of a total of five data fragments (3, 4, 5, 6, 7) and all fragments are stored consecutively, 4, 5, 6, 7, and 7 are accessed in order, and the contents of the file 1 are confirmed.

Or the other file (File 2) is composed of three data fragments 11, 25, and 17, and information of the metadata area can be used even if all fragments are discontinuously fragmented and stored. Confirm the contents by accessing the order of fragments.

However, in this case, except for the case where the two files (File 1, File 2) stored in FIG. 1 (a) are overwritten by overwriting the data area with an arbitrary value, The actual data of the storage area is maintained as it is, and if the attribute value of the metadata area is not initialized or changed, the corresponding information is searched to recover the file.

If the metadata region is lost as shown in FIG. 1 (c), since the location information of the fragments in which the file is stored can not be confirmed, the start fragment of the file is detected with respect to the entire storage region, Or to verify the internal format structure.

However, since such a file recovery method generally restores a normal file state only when the entire data fragment constituting the deleted file exists continuously, if the data fragment is discontinuously fragmented, the file recovery to the normal state is performed There was a difficult problem.

In particular, in the case of a moving image file, it is highly likely that a large number of fragments are discontinuously fragmented and stored, thereby making it difficult to recover the file if the file is deleted and the metadata area is lost. Recently, digital devices such as a smart phone, a tablet, a CCTV, and a black box have been widely used in connection with generation of a moving image file. In such a digital device, a storage medium based on a flash memory is used and used. In the case of such a flash memory, a fragmentation of data fragments may occur very much due to the nature of the storage mechanism, and a method of recovering a file to a normal state in the case where data fragments are discontinuously fragmented is required.

KR 10-2001-0064801 (Electronic document backup / recovery device and its method, Korea Telecommunication Corporation) 2001.07.11.

In order to solve the problems of the prior art as described above, the present invention classifies data fragments related to data to be restored, extracts image frames by reassembling data fragments based on the classification result, Which can be easily restored even when the fragment is fragmented into a plurality of data fragments.

According to an aspect of the present invention, there is provided a method for restoring data using data fragment classification, the method comprising: receiving restoration target data to be restored by an input unit; Classifying fragments related to the data to be restored among the previously stored data fragments in the storage medium; Reassembling the discontinuously fragmented fragments of the sorted fragments; And extracting an image frame for the restoration object data based on a result of reassembling the fragments of the image frame extracting unit.

In particular, the method may further include receiving input data of an input unit further including setting a size of a reference fragment for classifying the fragment.

In particular, the size of the reference fragment, which is a storage unit corresponding to at least one of a sector, a cluster, a block, and a page, may be included according to the type of the storage medium.

More preferably, the method further comprises a step of identifying a type of a fragment to be classified through a search of the image frame of the restoration target data in units of bytes. Searching for a fragment associated with the data to be restored in the data fragment stored in the storage medium based on the type of the identified fragment; And generating a fragment map for the restoration object data based on the retrieved fragment; And a step of classifying the fragment related to the restoration object data.

More preferably, the method further includes a step of removing overlapping fragments in the restoration object data based on the hash algorithm before performing the process of identifying the type of the fragment, And sorting.

Particularly, it is preferable that the type of the fragments to be classified through the search of the image frame of the restoration object data, which is at least one of key frame, delta-frame, decoding information and random, And identifying the type.

Further comprising classifying the fragment containing the beginning of the video frame based on the meta information stored in the header area of the key frame or delta frame if the type of the fragment is a key frame or a delta frame, And identifying the type of the fragment to be classified through the search of the image frame of the object data in byte units.

More preferably, receiving a fragment map for the restoration object data; Verifying validity of meta information stored in a header area of the restoration target data; A step of verifying a structure of an image frame of the restoration target data; And reassembling the image frame chain based on the validity of the fragment information and the verification result of the structure of the image frame. May include reassembling the discontinuously fragmented debris of the sorted debris.

More preferably, searching for meta information of the image frame chain exists; Acquiring the meta information when the meta information of the image frame chain exists; Setting a name of an image frame to be extracted; Extracting an image frame of an image file type based on the obtained meta information; And extracting an image frame for the restoration object data based on a result of reassembling the fragments.

And extracting image frames in the form of a moving picture file by arranging the image files in a time series order, extracting image frames of the restoration object data based on a result of reassembling the fragments .

More preferably, decoding the data of the image frame chain and extracting an image frame of an image file type when meta information about the image frame chain does not exist; Classifying at least one image set by applying an image processing technique to the extracted image file; And extracting an image frame for the restoration object data based on a result of reassembling the fragments.

In particular, it may include restoration target data which is at least one of a file system unassigned area, an unused area, a unassigned area of a single file, an unused area, a damaged file, and a raw dump.

In particular, a restoration process including at least one of time, place, GPS information, camera number, speed, container meta information, NAL header information, slice header information, frame meta information (time, place, GPS information, camera number, And may include meta information of the target data.

The data reconstruction method using the data fragment classification according to the present invention classifies data fragments related to data fragments to be reconstructed and extracts image frames by reassembling the classified data fragments so that data fragments to be reconstructed are discontinuously fragmented The data can be easily restored.

1 is a diagram illustrating a method of recording data in a storage medium in a digital device according to the related art.
2 is a flowchart illustrating a data restoration method using data fragment classification according to an embodiment of the present invention.
FIG. 3 is a flowchart illustrating a detailed procedure of a fragment classification step related to the restoration object data of the present invention.
Fig. 4 is a diagram showing an example of fragment classification stored continuously and discontinuously.
5 is a diagram illustrating an example of an H.264 video frame structure included in an MP4 container file.
6 is a diagram showing a result of data fragment classification.
7 is a flowchart showing a detailed procedure of the image frame chain reassembling step.
8 is a diagram illustrating a structure of an image frame chain using a single slice and multiple slices.
Figure 9 is a schematic representation of a verification scheme for reassembly of an image frame chain.
10 is a diagram illustrating an algorithm of a meta information verification process among verification schemes for reassembling the image frame chain.
11 is a diagram illustrating a verification of a video data structure in a verification scheme for reassembling a video frame chain.
12 is a view illustrating a process of reassembling an image frame chain using a single slice and multiple slices.
13 is a flowchart illustrating a detailed process of extracting an image frame for restoration target data based on a result of reassembling fragments of the image frame extracting unit.
14 is a diagram illustrating a meta information management type related to an image frame.
15 is a diagram showing a type of storing meta information in an image file.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT Hereinafter, the present invention will be described in detail with reference to preferred embodiments and accompanying drawings, which will be easily understood by those skilled in the art. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein.

Hereinafter, a data restoration method using the data fragment classification according to the present invention will be described in detail with reference to FIG.

2 is a flowchart illustrating a data restoration method using data fragment classification according to an embodiment of the present invention.

As shown in FIG. 2, the data restoration method using the data fragment classification according to the present invention is implemented through hardware means such as an input / output device, a processor, and a storage medium. First, (S110). At this time, the restoration target data to which the input unit is input is at least one of an unassigned area, an unused area, a unassigned area, a unused area, a damaged file, and a raw dump of a file system, . ≪ / RTI >

Particularly, in the file system, a unit of a specific size is used to store data in a storage medium. In this case, a specific size unit used may be referred to as a data fragment.

In addition, the size of the reference fragment may be set for the classification of the data fragment of the input unit. In this case, the size of the reference fragment to be set may be a sector, a cluster, a block, or a page, and represents a minimum storage unit. Thus, by setting the size of the data fragment to the minimum storage unit, the boundary where the fragmentation occurs can be quickly and easily grasped even if the data is stored in a fragmented state.

In addition, when there is decoding information of a previously reserved image frame, the decoding information is input as a default value, so that data can be restored even if decoding information can not be retrieved from the stream of the restoration object data.

Then, the fragment classifier, which is driven in the processor, classifies fragments related to the previously input restoration object data among the previously stored data fragments in the storage medium (S120).

Particularly, it is possible to perform a process of removing duplicated fragments in the restoration target data based on the hash algorithm before performing the process of identifying the type of fragments, May be omitted.

FIG. 3 is a flowchart illustrating a detailed procedure of a fragment classification step related to the restoration object data of the present invention.

As shown in FIG. 3, the fragment classifier first identifies the type of fragment to be searched for in the byte unit search for the image frame of the restoration object data (S121). At this time, the type of the identified fragment may be at least one of a key-frame, a delta-frame, decoding information, and random.

First, the key frame is a frame for storing an entire image, and no other frame is required for decoding. The key frame can be represented by an I-frame (Intra-coded frame) or an IDR (Instantaneous Decoding Refresh) frame according to a video codec .

Unlike the key frame, the delta frame requires a previous or later frame to be decoded, and a P-frame (Predicted frame), a B-frame (Bi-predictive frame) Frame. In particular, meta information of each frame is stored in a header area of the key frame or the delta frame. The fragments including the beginning part of the image frame can be classified using the meta information of each frame. For example, in the case of the H.264 video codec, a network abstraction layer (NAL) header and a slice header are included in the header area as meta information of a frame.

Next, the decoding information represents setting information that is indispensably required to decompress the data of the compressed and stored image frame to obtain visible information. Accordingly, if any fragmented decoding information is included, the fragment is classified into fragments including decoding information, and utilized for decoding of the image. For example, in the case of the H.264 video codec, the decoded information is stored as SPS (Sequence Parameter Set) and PPS (Picture Parameter Set) information.

Fourth, a random fragment corresponds to a case in which a data stream without a specific signature value has a high randomness. Generally, a video codec compresses and stores an image frame. Therefore, the data stream has high randomness. In the case where a key frame or a delta frame is large, an intermediate region of the image data can be stored in one or more random fragments.

In operation S122, it is determined whether the data fragment related to the input restore object data exists in the stored data fragment in the storage medium based on the type of the extracted fragment.

Hereinafter, a case in which fragments of H.264 image data included in the MP4 container file are continuously or discontinuously stored will be described in detail as an example.

4 (a) is a diagram showing an example of data fragments continuously stored.

As shown in FIG. 4 (a), there are some differences according to the method of storing video data, but generally, decoding information is separately managed and stored in a form in which video frames are continuously present. As a result of identifying the type of data fragment for the stored image frame, decoding information of H.264 was found in the first fragment, and the beginning of the key frame was found at the middle point of the second fragment. Also, fragment # 3 was randomly classified because no known signature value was found and the data stream was highly random. In addition, the beginning of the delta frame was found in the fourth fragment, and the fifth fragment was randomly classified as high in water.

On the other hand, let us examine the case where fragments of H.264 image data contained in the MP4 container file are stored discontinuously and fragmented.

As shown in Fig. 4 (b), by identifying the type of the data fragment, the beginning of the delta frame was found in the first fragment, and the beginning of the key frame was found in the middle of the second fragment. In addition, H.264 decoding information was found in the third fragment, and the fourth and fifth fragments were randomly classified because the known signature value was not found and the data stream was highly random.

Particularly, header fragments of key frames and delta frames can be classified using header information of image frames. At this time, there is a difference in the kinds of header information available according to the container file (MP4, AVI, MKV, etc.) in which data of the image frame is stored or the video codec (H.264, MPEG-4 visual, etc.) Can be used to effectively detect the beginning of an image frame, and can also be useful in the process of reassembling fragmented data fragments.

For example, in the case of image data generated by the H.264 video codec, a header field includes a network abstraction layer (NAL) header and a slice header.

In this case, the NAL is a unit used for storing data in the H.264 codec, and the decoding information and each frame (or pictures) exist in an independent NAL structure.

Table 1 is a table showing various types of NAL.

NAL type Explanation NAL header value
(8 bits) One Coded slice of a non-IDR picture Delta-frame 0x01, 0x21, 0x41, 0x61 5 Coded slice of an IDR picture Key-frame 0x25, 0x45, 0x65 6 Supplemental enhancement information Additional Information 0x06 7 Sequence parameter set (SPS) Decoding information 0x27, 0x47, 0x67 8 Picture parameter set (PPS) Decoding information 0x28, 0x48, 0x68

Key-frame, delta-frame and decoding information (SPS, PPS) are basically used as shown in Table 1 above.

Slice represents a storage unit of image data stored in one NAL structure, and is a component constituting a picture (Picture, visible image) in a video codec. Generally, pictures and slices are composed of 1: 1, and one picture is included in one NAL structure. However, if one picture is composed of a plurality of slices, only a part of the picture is included in the slice stored in one NAL structure.

Table 2 shows the slice type values used in the H.264 video codec.

Slice type Explanation 0, 1, 3, 5, 6, 8 Delta-slice 2, 4, 7, 9 Key-slice

As shown in Table 2, the value of the slice type used in the H.264 video codec is divided into a key slice or a delta slice. At this time, since the slice type used is included in the NAL structure, if the image data is normal, the NAL type and the slice type must coincide with each other.

5 is a diagram illustrating a structure of an H.264 video frame included in an MP4 container file.

5, when the H.264 data is included in the MP4 container file, since the frame length information of the corresponding NAL is 4 bytes long at the beginning of each NAL structure, It can be utilized as information.

In particular, since the length of one NAL structure varies depending on the data and encoding settings, it is usually within 200 kilobytes (KB) in 320p, 480p, 720p, and 1080p, The accuracy of the classification can be increased.

After 4 bytes of length information, the NAL header and the slice header are immediately located. The header field is interpreted according to the grammar defined in the standard to obtain and verify the value of each field. Each field in the NAL structure is divided in units of bits. U (count) in FIG. 5 indicates reading of a count of bits, ue (v) means Exp-Golomb code, Read variable length bits. In this case, count represents the number of bits 0 before the bit 1 is outputted, and bit 0 is removed by the count number, and the ue (v) can be defined as Equation (1).

[Equation 1]

u (v) = u (count + 1) - 1

The value of the start position (first_mb_in_slice) of the macroblock existing in the slice header is determined by the relation between the picture and the slice (0 if the relation is 1: 1), and the meaning of the slice type (slice_type) can be confirmed through Table 2. The frame number field (frame_num) calculated by the value (log2_max_frame_num_minus4) obtained through the decoding information indicates the current frame number. The key frame is 0, and the delta frames existing until the next key frame are sequentially incremented Are assigned.

In this case, each of the above-described field values must have a valid range, and since the slice is included in the NAL structure, if the NAL type is a key frame, the slice type must pass the key slice to pass the verification.

5 (a) shows H.264 image frames continuously stored in the MP4 container file, and shows important information utilized in the fragment classification step. In the fragment classification, when the [frame length] [NAL header] [slice header] area is normally detected when a byte unit search is performed on the input fragment as shown in FIG. 5 (a), it is classified into image frame fragments.

On the other hand, if the H.264 video data is not included in the MP4 container file but exists in the H.264 raw storage format, the syncword (0x00000001) instead of the length information is placed in front of each NAL structure. Therefore, even in the case of the row storage format, it is possible to detect effectively by checking the syntax of syncword and NAL / slice header.

In addition, fields such as a NAL type (nal_unit_type), a slice type (slice_type), a frame number (frame_num), and a start position (first_mb_in_slice) of a macroblock are utilized in a subsequent reassembling step.

In particular, H.264 decoding information is stored in the SPS and PPS NAL structures. 5B is a detailed view of the H.264 decoding information area included in the MP4 container file, and the storage structure of SPS and PPS NAL can be confirmed. As can be seen in FIG. 5 (b), the grammar verification of the avcC item managing the H.264 decoding information in the MP4 container can be very effectively detected by searching the structure in which the PPS is continuously present after the SPS.

In addition, the SPS NAL consists of a 1-byte NAL header and decoding information. In particular, there is a log2_max_frame_num_minus4 field for checking the number of a video frame. The PPS NAL consists of a 1-byte NAL header and decoding information.

In addition, since the image data generated by the same encoding setting has the same decoding information (SPS, PPS in the case of H.264), even if decoding information is not found in the entire classification target data fragment, the user can define it and utilize it for decoding.

Referring back to FIG. 3, a fragment map for the restoration object data is generated based on the retrieved fragment (S123).

6 (a) and 6 (b) illustrate the result of classification of the data stored in the storage medium. FIG. 6 (a) . Unknown fragments that do not have any special signature information such as key fragments, delta frames, decoding information, and randomness are used as key elements in subsequent reassembly processes. Accuracy can be increased.

Referring back to FIG. 2, the fragment reassembly unit driven by the processor reconstructs discontinuously fragmented fragments of the classified fragments to generate an image frame chain (S130).

7 is a flowchart showing a detailed procedure of the image frame chain reassembling step.

As shown in Fig. 7, the debris remapping unit receives the debris map generated from the debris classifying unit for the restoration object data (S131).

Then, the validity of the meta information stored in the header area of the restoration target data is verified (S132). At this time, the meta information of the restoration target data may include at least one of container meta information, NAL header information, slice header information, and frame meta information. At this time, the frame meta information may include at least one of time, place, GPS information, camera number, and speed.

Next, the structure of the video frame of the restoration target data is verified (S133).

The image frame chain is generated based on the validity of the fragment meta information and the verification result of the structure of the image frame (S134). In this case, the generated image frame chain can be defined as shown in Table 3 below.

Frame chain (1) = key-frame [0] | Delta-frames [1] | Delta-frame
[2] ... delta-frame [N-1]
Frame chain (2) = key-frame [0] | Delta-frames [1] | Delta-frame
[2] ... delta-frame [N-1]
...
Frame chain (M) = key-frame [0] | Delta-frames [1] | Delta-frame [2] ... delta-frame [N-1]
Assumption 1: Average frame rate of image data = N frames / sec
Assumption 2: Picture: Slice = 1: 1 (the value of the start position of the macroblock in the slice header is always 0)
* Symbol Meaning: (Chain No.), [Frame No.]

In the reassembling step of the image frame chain, the image frame chain is constructed by reassembling all possible fragments for each key frame obtained through the classification step described above. At this time, in order to perform the reassembly more efficiently, a process of constructing a separate fragment map (FIG. 6 (b)) is performed by only the fragment related to the image frame in the entire fragment map (FIG. 6 .

Hereinafter, the concept of the image frame chain using single and multiple slices will be briefly described before reassembling the image frame chain for each key frame.

8 (a) shows a structure of a video frame chain using a single slice.

As shown in Fig. 8 (a), since the picture and the slice have a 1: 1 relationship, it can be seen that the frame and the slice also have a 1: 1 correspondence. Thus, as each picture is stored in one frame (= slice), the macroblock start position value (first_mb_in_slice, abbreviated as first_mb) is always zero. The frame number (frame_num) has a value (1 or more) sequentially increasing from the key-frame (0) to the delta frame before the next key frame. 8A, since the average frame rate of the image data is 30 frames per second, the delta frames are numbered up to 29 times on average.

Next, an image frame chain structure using multiple slices will be described.

8 (b) is a diagram showing a structure of an image frame chain using multiple slices.

As shown in Fig. 8 (b), since a picture and a slice have a 1: N relationship, it can be seen that a frame and a slice also have a 1: N form. Accordingly, as each picture is divided into a plurality of slices and stored, the macroblock start position value (first_mb) of each slice is sequentially increased in one frame. The frame number (frame_num) has a value (1 or more) that sequentially increases from a (0) to a delta frame before the next key frame. Since one frame is composed of a plurality of slices, Are all given the same frame number.

Depending on the single slice or multi-slice environment, the complexity of the reassembly process of the image frame chain can vary in some ways, but the validity of the fragments can be verified using the key features described above.

Such image frame chain reassembly is performed based on each key frame.

A verification scheme for reassembling the image frame chain based on one H.264 key frame can be schematically shown in FIG.

In the verification step for the reassembly of the image frame chain, first, meta information verification is performed, and then, image data structure verification is performed.

10 is a diagram illustrating an algorithm of a meta information verification process among verification schemes for reassembling the image frame chain.

As shown in FIG. 10, in the verification process, validity of various meta information such as container meta information, NAL header information, and slice header information is verified.

The first verification process uses meta information. Therefore, if only the fragment having the beginning of each image frame is performed, the structure verification process of the image data verifies the image data structure for all the candidate data fragments.

11 is a diagram illustrating a verification of a video data structure in a verification scheme for reassembling a video frame chain.

As shown in FIG. 11, the verification of the video data structure is performed through a grammar check in a process of decoding a compressed image. If it is a valid data fragment, all byte streams input to the decoder are processed, but if it is an invalid data fragment, an error occurs in the middle.

For example, in FIG. 9, the first fragment of the image frame chain is identified through the verification of the meta information and the structure verification of the image data as a result of the verification of the beginning of the key frame existing in the first fragment. Also, the second and third fragments were only validated for the structural verification of the image data as the middle part of the image data. On the other hand, the 4th fragment contains the beginning of the delta frame but fails to pass the first meta information verification and is not included in the image frame chain. The 5th fragment is validated and included in the chain. The validation of the meta information and the verification of the image data structure can be performed effectively and the reassembly of the image frame can be performed.

12 (a) shows a process of reassembling a video frame chain using a single slice.

As shown in Fig. 12 (a), based on the result of the data fragment classification, an attempt is made to reassemble the image frame chain based on the fragment 2 which is the beginning of the key frame. As a result of the meta information verification for fragment 2, it can be confirmed that the frame number and the start position of the macroblock are 0, and the length of the entire frame is 30α. Since the length of the frame included in the current fragment is 9α, Is required. At this time, 21α required for reassembly is longer than one fragment (16α), so the next fragment should be a random fragment. First, we attempted to verify the image data structure for fragment 3, but an error occurred in the intermediate process. Next, fragment 4 successfully passed the verification for normal 16α data. Also, since the end portion 5α of the key frame needs to be reassembled after the fragment 4, it is tried to verify the next candidate fragments so that a fragment having a delta frame start position after 5α, a frame number of 1, 1 < / RTI > In addition, the delta frame with frame number 1 included in fragment 1 has a length of 27 alpha, and since the length of the frame included in the current fragment is 11 alpha, reassembly for the remaining 16 alpha is required. At this time, since 16α which is required to be reassembled is longer than one fragment, the following fragment should be a random fragment, and normal decoding of all the fragment 3 data has succeeded and passed verification. In this way, all the delta frames can be reassembled based on the respective key frames through the above-described method, thereby constructing the image frame chain.

12 (b) is a view illustrating a process of reassembling the image frame chain using multiple slices.

As shown in Fig. 12 (b), since a frame and a slice have a 1: 2 relationship, a slice having the same frame number and a macroblock start position of 0 or more may exist. 12 (a), a detailed description thereof will be omitted.

Returning to FIG. 2, the image frame extraction unit extracts the image frame for the restoration object data based on the result of reassembling the fragments (S140).

13 is a flowchart illustrating a detailed process of extracting an image frame for restoration target data based on a result of reassembling fragments of the image frame extracting unit.

As shown in FIG. 13, first, it is determined whether meta information about the image frame chain exists (S141).

If meta information for the image frame chain exists, the meta information is obtained (S142).

Next, the name of an image frame to be extracted is set (S143).

And extracts image frames in the form of an image file based on the meta information obtained in step S144.

In addition, the extracted image files are listed in a time series order to generate a moving image file (S145).

If there is no meta information for the image frame chain in step S141, the image data of the image frame chain is decoded to extract an image frame in step S146.

Thereafter, the image processing method is applied to the extracted image file to classify at least one image set having similar contents (S147).

Through this image frame extraction process, meaningful information can be obtained from a single fragment or a reassembled image frame chain, and convenient and intuitive confirmation is possible.

As described above, when the image frame fragments are successfully reassembled, the contents of each image frame can be visually confirmed. However, if there are thousands to tens of thousands of image frames in the target data stream, it may take a very long time to check them one by one. Hereinafter, a method of effectively extracting each image frame will be described in detail.

In the process of extracting an image frame, first, the existence of meta information related to a single debris or a reassembled image frame chain is checked. Examples of the meta information related to the image frame include a camera number, a photographing time, a photographing place, and other additional information, and there are various methods of managing such meta information by each image data generating apparatus.

14 is a diagram illustrating a meta information management type related to an image frame.

A method of managing meta information of an image frame is divided into two types. As shown in FIG. 14 (a), meta information can be obtained by analyzing the storage structure, and meta information of each image frame is separately managed However, when image data is included in a container file such as MP4, it is general not to store meta information separately. However, embedded video recording systems such as CCTV, automobile black box, etc., may include meta information in the form of text in the image file itself for monitoring and recording purposes. FIG. 14B shows an example of a type in which meta information is included in an image file. Meta information in an image file can be extracted using optical character recognition (OCR) Information such as information, camera number, speed, etc. is mainly stored.

If there is meta information about the image frame, the name of the corresponding image frame is determined using the meta information. The name of the image frame is determined in the form of a photographing time, a camera number, and the like, and then extracted into a visible image file so that the image frames can be continuously checked in time even if they are fragmented and discontinuous.

In addition, it is possible not only to acquire continuous images temporally through the name setting of image frames, but also to reconfigure a plurality of images into a single playable video file, thereby enhancing the convenience and intuitiveness of content confirmation.

On the other hand, when there is no meta information about the image frame, since there is no criterion for determining the name, only extraction into a visible image file is performed without setting a name. Although the extracted image files can be visually checked, it is possible to improve efficiency by classifying the image into a similar image group using image processing.

Although the present invention has been described with reference to the image data generated by the H.264 video codec, the image data generated by various CODECs such as MPEG-4 visual, HEVC, Assembly, and extraction steps of the present invention.

In addition, the data restoration method using the data fragment classification may be stored in a computer-readable recording medium on which a program for execution by a computer is recorded. At this time, the computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. Examples of the computer readable recording medium include ROM, RAM, CD-ROM, DVD 占 ROM, DVD-RAM, magnetic tape, floppy disk, hard disk, optical data storage, and the like. In addition, the computer-readable recording medium may be distributed to network-connected computer devices so that computer-readable codes can be stored and executed in a distributed manner.

The data restoration method using the data fragment classification according to the present invention classifies data fragments related to the data fragments to be restored and extracts the image frames by reassembling the classified data fragments so that the data fragments to be restored are discontinuously fragmented The data can be easily restored.

While the present invention has been described in connection with what is presently considered to be practical exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, Do.

Claims (14)

Receiving input data to be restored by the input unit;
Classifying fragments related to the data to be restored among the previously stored data fragments in the storage medium;
Reassembling the fragments which are discontinuously fragmented among the fragments classified by the fragment reassembling section using the meta information of the restoration object data and the verification of the frame structure; And
Extracting an image frame for the restoration object data based on a result of reassembling the fragments of the image frame extracting unit;
Wherein the data fragment classification includes at least one of the following:
The method according to claim 1,
The step of receiving the input unit data
Further comprising setting a size of a reference fragment for classifying the fragment.
3. The method of claim 2,
The size of the reference fragment is
Wherein the storage unit is a storage unit corresponding to at least one of a sector, a cluster, a block, and a page according to a type of the storage medium.
The method according to claim 1,
The step of classifying the fragment related to the restoration object data by the fragment classifier
Identifying a type of a fragment to be classified through a search for a video frame of the restoration target data in byte units;
Searching for a fragment associated with the data to be restored in the data fragment stored in the storage medium based on the type of the identified fragment; And
Generating a fragment map for the restoration object data based on the retrieved fragment;
The method of claim 1, further comprising:
5. The method of claim 4,
The step of classifying the fragment related to the restoration object data by the fragment classifier
Removing the duplicated fragments in the restoration object data based on the hash algorithm before performing the process of identifying the type of the fragment;
The method of claim 1, further comprising:
5. The method of claim 4,
The process of identifying the type of fragment to be classified through the search of the image frame of the restoration target data in byte units
Wherein the type of the fragment is at least one of a key-frame, a delta-frame, decoding information, and random.
The method according to claim 6,
The process of identifying the type of fragment to be classified through the search of the image frame of the restoration target data in byte units
Further comprising classifying fragments including a beginning portion of an image frame based on meta information stored in a header region of the key frame or delta frame if the type of the fragment is a key frame or a delta frame Data Reconstruction Method Using Fragment Classification.
8. The method of claim 7,
Wherein the step of reassembling the fragmentally fragmented fragments among the fragments classified by the fragment reassembling section using the meta information of the restoration object data and the verification of the frame structure
Receiving a fragment map for the restoration object data;
Verifying validity of meta information stored in a header area of the restoration target data;
A step of verifying a structure of an image frame of the restoration target data; And
Reassembling the image frame chain based on the validity of the fragmentation meta information and the verification result of the structure of the image frame;
The method of claim 1, further comprising:
9. The method of claim 8,
The step of extracting an image frame for the restoration target data based on a result of the image frame extracting unit reassembling the fragments
Searching whether the meta information of the image frame chain exists;
Acquiring the meta information when the meta information of the image frame chain exists;
Setting a name of an image frame to be extracted; And
Extracting an image frame of an image file type based on the acquired meta information;
The method of claim 1, further comprising:
10. The method of claim 9,
The step of extracting an image frame for the restoration target data based on a result of the image frame extracting unit reassembling the fragments
Extracting image frames in the form of a moving picture file by arranging the image files in a time series order;
The method of claim 1, further comprising:
10. The method of claim 9,
The step of extracting an image frame for the restoration target data based on a result of the image frame extracting unit reassembling the fragments
Decoding the data of the image frame chain and extracting an image frame of an image file type when meta information about the image frame chain does not exist;
Classifying at least one image set by applying an image processing technique to the extracted image file;
The method of claim 1, further comprising:
The method according to claim 1,
The restoration object data
Wherein the data fragment is at least one of an unassigned area, an unused area, a unassigned area, a unused area, a damaged file, and a raw dump of a file system.
The method according to claim 1,
The restoration object data
Container meta information, NAL header information, slice header information, frame meta information Wherein the data fragment classification includes at least one of the following:
A computer-readable recording medium on which a program for executing a method according to any one of claims 1 to 13 is recorded.

Images