JP2011234068A - Video recording device and video recovery device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a mechanism which realizes purposes of accelerating a process of completion of video recording and speedy recovery of index information in occurrence of a fault.SOLUTION: A video recording device comprises: an encoding unit 601 which encodes an externally inputted video signal by an interframe compression system; a video writing unit 602 for recording the video data encoded by the encoding unit 601 on a recording medium 610 as a video file 611; an index recording unit 603 which prepares index information of the video data encoded by the encoding unit 601 and records the same on the recording medium 610 as an index file 612; and an index composition unit 604 which integrally composes the video file 611 and the index file 612 after the video file 611 and the index file 612 are recorded on the recording medium 610.

Description

  The present invention relates to a video recording apparatus that records video and a video restoration apparatus that performs restoration processing of the recorded video.

  Conventionally, program production at a broadcasting station has recorded video on a tape using a camera recorder, and edited and archived using the tape. In recent years, camera recorders that record video on a randomly accessible recording medium such as an HDD or a flash memory have appeared. Accordingly, a style has been established in which video data is recorded in a file format with a camera recorder, and the file is edited with a non-linear editing machine.

  In order to record a file on a recording medium, a module called a file system for managing files on the recording medium is used. A typical file system is FAT (Standard ECMA-107: Volume and File Structure of Disk Cartridges for Information Interchange). Since the operation of the file system is complicated, it is common to mount an operating system (OS) on the camera recorder and implement the file system as software.

  As an example of the file system format, the FAT format will be described with reference to FIG. As shown in FIG. 1, a physical drive 100 such as a memory card or a hard disk is generally composed of a master boot record 102 and one or more logical drives 101 placed at the head of the area. One logical drive 101 is formatted with one type of file system. In the case of the FAT format, a system area such as a partition boot sector 103, a file allocation table (FAT) 104, a FAT backup 104b, and a root directory entry 105 is arranged from the top of the area, and a user data area 106 is arranged behind the system area. The In the partition boot sector 103, information necessary for starting the partition such as the number of sectors of the partition is recorded. File recording information is arranged in the FAT 104. In the user data area 106, file data itself is placed. Files are recorded in units called clusters (usually 4 KB to 32 KB).

  In the root directory entry 105, information on files or folders placed in the root directory is recorded as entries. In the hierarchical directory, all files and directories are recorded as a directory entry 107 consisting of 32 bytes as shown in FIG. In FIG. 2, FLOWER02. A directory entry of a file having a file name of AVI is shown. Thus, in the FAT format, a file is recorded in an 8.3 format with a file name of 8 characters and an extension of 3 characters. The directory entry 107 includes the first cluster number 108 in which file data is placed. Access to the file data is enabled by this first cluster number 108.

  The recording information of the file arranged in the FAT 104 will be described with reference to FIG. The FAT 104 is a table for managing clusters on the user data area 106. The FAT 104 records the cluster number in which the next data of the file exists in units of 12 bits or 16 bits. As shown in FIG. 3, when the file 120 to be recorded in the user data area 106 is divided into partial data 121 and recorded in clusters with cluster numbers 3, 4, and 7, the cluster number is stored in the FAT 104 as information for connecting the clusters 4 is recorded at the position of 3, and 7 is recorded at the position of the cluster number 4. Finally, an end mark is recorded at the position of cluster number 7. The leading cluster number 3 in the file 120 is recorded in the directory entry 107 of the file 120 as described above.

  When a video file is recorded on a recording medium using a camera recorder, the video signal is generally compressed and recorded from the viewpoint of capacity. The MPEG2 (Moving Picture Experts group 2) standard is known as one of the typical standards for video data compression coding. MPEG video data is composed of units called GOP (Group of Pictures), which is a collection of several frames of data.

  One GOP includes image data generated by compressing and encoding a predetermined number of original images. Each image data is classified into three picture types, an I picture (Intra-coded picture), a P picture (Predictive-coded picture), and a B picture (Directive-predictive-coded picture) according to the compression coding method of the original picture. .

  An I picture is also called an intra-coded image, and is coded using only information of one original image. Therefore, the I picture is independent of the other image data pictures before and after it and can be decoded only by the I picture itself.

  The P picture is also called an interframe forward predictive encoded image, and uses a previously decoded I picture or P picture temporally before as a predicted image (an image serving as a reference for obtaining a difference), A difference from the motion compensated predicted image is encoded.

  The B picture is also called a bi-predictive encoded image, and is encoded as an image inserted between the I picture and the P picture after being processed first. That is, the B picture is a predicted image of an already decoded I picture or P picture earlier in time, an already decoded I picture or P picture later in time, and an interpolated image made from both. Three types are used.

  A technique is disclosed that allows direct access to a GOP by creating the GOP address as an index file in MPEG2 stream data (see Patent Document 1). Thereby, it is possible to jump to the head of an arbitrary GOP. However, this method has a problem that management is complicated because MPEG2 stream data and index files must be managed in association with each other.

  A technique is disclosed in which random access can be realized without having index information by setting the data length of one frame to a fixed length (see Patent Document 2). However, since unnecessary data is inserted, useless data areas are consumed, and it is not an efficient method especially for a removable disk or the like having no abundant data areas.

  In order to realize random access for each frame, a file format has been proposed in which index information for each frame can be arranged in the same file as video data. One of them is MXF (Material Exchange Format, SMPTE 377M). MXF is a file format mainly for handling professional-use digital video and audio. When recording video data with MXF with a camera recorder, the recording time cannot be determined before shooting, and therefore the data size of the index information cannot be determined in advance. In such a case, the index information is generally written not at the beginning (header) but at the end (footer) of the file.

JP-T-2006-524410 JP 2008-124931 A

  However, if a large amount of data is written at the end, there is a problem that it takes time to stop recording and it is not possible to quickly enter the next recording. In addition, when the recording is not normally completed, such as when the power is turned off halfway, the index information does not remain on the recording medium. To recover an abnormal file, it is necessary to recreate index information from the video data. This process needs to read a large amount of video data, and thus it takes a lot of time to complete.

  In view of the above-mentioned problems, the present invention is to increase the speed of the recording end process while recording the index information and video data for each frame on the same file, and to quickly restore the index information when a failure occurs. It aims at providing the mechanism which realizes one.

  The video recording apparatus of the present invention includes an encoding unit that encodes an externally input video signal using an inter-frame compression method, a video writing unit that records video data encoded by the encoding unit on a recording medium as a video file, An index recording unit for creating index information of the video data encoded by the encoding unit and recording the index information on the recording medium as an index file; and after recording the video file and the index file on the recording medium, the video file And an index composition unit for synthesizing the index files into one.

  By combining the index information with the video file by operating the file system at the end of recording, writing to the file can be suppressed compared to adding index information to the video file. Can finish. Further, when a failure occurs during recording and the recording is not completed normally, the repair process can be proceeded by relying on the index file, so that the repair process can be completed in a short time.

Diagram showing FAT format Diagram showing directory entry The figure about the record information of the file arranged in FAT Diagram explaining MXF file format The figure explaining the index information of MXF FIG. 2 is a block diagram illustrating a configuration of a video recording apparatus according to Embodiment 1. Flowchart for recording video of Embodiment 1 Flowchart for synthesizing indexes according to the first embodiment The figure which shows FAT before and after synthesize | combining the index of Embodiment 1. The figure which shows MXF created in Embodiment 1 FIG. 3 is a block diagram illustrating a configuration of a video restoration device according to the first embodiment. The figure which shows the structure of data when there are few video files than the number of frames of the index file in Embodiment 1. The figure which shows the structure of data when there are more video files than the number of frames of the index file in Embodiment 1. FIG. 3 is a block diagram showing a configuration of a modification of the video restoration device according to the first embodiment. Flowchart of video file restoration according to the first embodiment Diagram showing the internal structure of an MPEG picture Flow chart of index creation according to Embodiment 1

(Embodiment 1)
In the present embodiment, the video recording device 600 and the video restoration device 1100 will be described by taking MXF as an example of the file format of the video file. MXF is composed of a collection of data elements called KLV (Key-Length-Value), and all data such as metadata and video data has a KLV structure. The KLV structure is a structure in which a key (Key), a length (Length), and a value (Value) are sequentially arranged from the top, and what data is arranged in the value in the key. A 16-byte label that conforms to the SMPTE 298M standard is placed. In the length, the data length (8 bytes) of data arranged in the value is arranged by BER (Basic Encoding Rules: ISO / IEC 882-1 ASN). Actual data is placed in the value.

  FIG. 4 shows the MXF file format. A file header (File Header), a file body (File Body), and a file footer (File Footer) are sequentially arranged from the top.

  The file header is configured by sequentially arranging a header partition pack (Header Partition Pack) and header metadata (Header Metadata) from the top. In the header partition pack, a pattern for specifying a header, a format of data arranged in a file body, information indicating an MXF file format, header size information, and the like are arranged.

  The file body is configured by sequentially arranging a body partition pack and an essence container. Image data is arranged in the essence container. Since the essence container also has a KLV structure, image data is arranged as a value after the key and length.

  In the file footer, a footer partition pack (Footer Partition Pack) and an index table (Index Table) are sequentially arranged. The index table stores data indicating the position of image data, and will be described in detail later with reference to FIG.

  FIG. 5A shows the display order of each image. The numerical part indicates the display order, and the alphabet (I, B, P) indicates the picture type, indicating that the display is performed in the order of B1, I2, B3, P4, B5, P6. FIG. 5B shows how the video of FIG. 5A is stored on the file. In order to decode a B picture, it is necessary to decode an I picture or a P picture that is the predicted picture in advance. Therefore, images are recorded in an order different from the display order so that data can be read and decoded sequentially from the beginning of the file. For example, since B1 is decoded using I2, it is recorded on the file in the order of I2 and B1. Similarly, P4 is disposed forward of B3 and P6 is disposed forward of B5. Therefore, the recording order is I2, B1, P4, B3, P6, and B5.

  FIG. 5C shows details of the MXF index table. “Temporal Offset” represents conversion from display order to recording order. It can be seen that the frame B1 of the first frame is stored in the file as data of the second frame obtained by adding 1 of Temporal Offset to the frame number 1 of the display. Similarly, it can be seen that I2 is stored in the file as data of the first frame obtained by adding -1 of Temporal Offset to frame number 2 of the display. The key-frame offset is information indicating which frame needs to be decoded in order to decode the frame, and normally refers to an I picture. For example, in order to decode B3, it is necessary to decode from the second frame (I2) obtained by adding −1 of the Key-Frame Offset from display frame number 3. Flags is the picture type described above. The stream offset is information indicating in which position on the file the frame is stored. This information is recorded based on the recording order frame number. For example, when the byte offset in which B3 is stored is obtained, the frame number 4 of the recording obtained by adding 1 of Temporal Offset to frame number 3 in the display order is obtained, and the byte offset of 65,000 bytes is obtained from the stream offset of the fourth frame Ask. In order to output a frame at an arbitrary position, it is only necessary to decode from the key frame to the output image. For example, in order to output B3, −1 of the Key-Frame Offset is added to the frame number 3 of the display to obtain the display frame number 2 (I2), and the Temporal Offset and Stream Offset of I2 are obtained as described above. Byte offset 0 is obtained. Next, the byte offset 65,000 bytes of B3 is obtained from the Temporal Offset and the Stream Offset. By reading the picture from the 0th byte to the 65,000th byte from the information of these two byte offsets, decoding can be performed in the order of I2, B1, P4, and B3, and B3 can be output.

  FIG. 6 shows the configuration of the video recording apparatus 600 of the present embodiment. The encoding unit 601 performs inter-frame compression on an externally input video signal, and encodes (encodes) it into MPEG format compressed data. The video writing unit 602 records the video data encoded by the encoding unit 601 as a video file 611 on the recording medium 610. The index recording unit 603 records the index information of the compressed data encoded by the encoding unit 601 as an index file 612 on the recording medium 610. The index composition unit 604 connects the index file 612 to the end of the video file 611 when the recording is finished.

  The processing flow of the video recording apparatus 600 will be described with reference to the flowchart of FIG. 7 and FIG. The encoding unit 601 first receives the video signal of the first frame in the data input process S71. Since the video signal of the first frame is scheduled to be a B picture, the encoding unit 601 decides to delay the encoding of the video signal of the first frame in encoding propriety determination S72, and uncompressed image In the storage process S73, it is temporarily stored in the uncompressed image holding unit D70. The encoding unit 601 holds an uncompressed image holding unit D70 inside. Next, when the video signal of the second frame is received in the data input process S71, since the video signal of the second frame is scheduled to be an I picture, the encoding unit 601 determines in the encoding availability determination S72. It is determined to encode the video signal of the second frame. In the encoding process S74, the encoding unit 601 first encodes the video signal of the second frame, then extracts the video signal of the first frame stored in the uncompressed image holding unit D70, and outputs the video signal of the second frame. Is used to perform inter-frame compression. In the data writing process S75, the video writing unit 602 writes the MXF file header, body partition pack, and K and L data of the essence container shown in FIG. 4 to the video file 611 before writing the compressed data. Next, the video writing unit 602 writes the compressed data (I2) of the second frame and the compressed data (B1) of the first frame to the video file 611 in this order. Since the compressed data displayed in the second frame is written in the first frame on the video file 611, the Temporal Offset in the first frame is 1 (2-1). Since the I picture that is a key frame is in the second frame, the Key-Frame Offset of the first frame is 1 (2-1). In addition, Flags of the first frame is B, and Stream Offset is 0. Since the compressed data displayed in the first frame is written in the second frame on the video file 611, the Temporal Offset in the second frame is -1 (1-2). Since the I picture that is a key frame is in the second frame, the Key-Frame Offset of the second frame is 0 (2-2). If the data size of I2 is 50,000 bytes, Flags of the first frame is I, and Stream Offset is 50,000.

  In the index writing process S76, the index recording unit 603 writes these two frames of index information to the index file 612. Since recording is not stopped in two frames, it is determined in recording stop determination S78 that recording is not stopped. Next, in order to process the third frame, the process returns to the data input unit S71. When it is determined in the recording stop determination S78 that the recording is finished, the index composition unit 604 performs the index composition processing S79, and the recording is finished. The index composition process S79 is a process for composing the video file 611 and the index file 612 into one file.

  FIG. 8 shows a flowchart of the index composition process S79. The index composition unit 604 executes index composition processing S79. A description will be given assuming that the header size is 32 KB, the compressed data size is 82 KB, and the size of one cluster is 32 KB. In the data size calculation process S81, the header size 32KB and the compressed data size 82KB are added together to obtain the data size 114KB recorded in the video file 611. In the remaining cluster calculation processing S82, 114 KB = 3 clusters (96 KB) +18 KB bytes, so the remaining 14 KB (32 KB-18 KB) in the cluster is obtained. The filler embedding process S83 is a process for embedding a size obtained by subtracting the footer partition pack from the size calculated in the cluster remaining amount calculating process S82 as data for stuffing. As shown in FIG. 4, the filler also has a KLV structure, and data padded with 0s after K and L are arranged. Assuming that the footer partition pack is 1 KB for simplicity, in the filler embedding process S 83, a 13 KB filler is created by subtracting 1 KB of the footer partition pack from the 14 KB up to the end of the cluster, and is added to the video file 611.

  FIG. 9 shows the FAT of the video file 611 and the index file 612. As shown in FIG. 9A, it is assumed that video data is recorded in clusters with cluster numbers 2, 3, 4, and 5 and an index file 612 is recorded in cluster number 7. The index composition unit 604 rewrites the next cluster number of the cluster 5 from the end mark to 7 in the FAT operation process S84, and as shown in FIG. 9B, the index composition unit 604 creates the cluster numbers 2, 3, 4 5 and 7 are changed to constitute one file. Thereafter, the index composition unit 604 deletes the index file 612 from the directory entry, and completes the combination of the video file 611 and the index file 612. Thereby, the two files are combined into one correct MXF file at the end of recording, and a video file 611 as shown in FIG. 10 is completed.

  As described above, the video recording apparatus records the video data input from the outside using the inter-frame compression method and the video data encoded by the encoding unit 601 on the recording medium 610 as the video file 611. A video writing unit 602 that creates the index information of the video data encoded by the encoding unit 601, and records the index file 612 as an index file 612 on the recording medium 610. By providing an index composition unit 604 that synthesizes the video file 611 and the index file 612 after recording, the index information is collectively added to the video file at the end of recording. The scan file 612 for only synthesizes the video file 611, it is possible to finish the recording end process in a short time. Further, the video recording apparatus can proceed with the repair process by relying on the index file 612 when a failure occurs during the recording and the recording is not normally completed, so that the repair process is completed in a short time. be able to.

  Next, the video restoration apparatus of this embodiment will be described. This device is a device that restores a normal video file 611 from a state in which some trouble occurs during recording by the video recording device 600 and the video file 611 and the index file 612 remain halfway.

  First, FIG. 11 shows the configuration of the video restoration apparatus 1100 when there is no contradiction between the contents of the index file 612 and the contents of the compressed file. In the video restoration apparatus 1100, the same components as those of the video recording apparatus 600 are assigned the same numbers, and descriptions thereof are omitted. The data recovery instruction unit 1101 first confirms that the index file 612 and the video file 611 exist, and instructs the index composition unit 604 to combine the index file 612 with the end of the video file 611. The index composition unit 604 performs the processing as described above to complete the file as shown in FIG.

  The video recording apparatus 600 according to the present embodiment writes to a file via a file system. The file system has a mechanism called delayed writing. Delayed writing is a process of storing data in a memory once and writing to a recording medium after a data writing command of a certain size is issued, instead of writing to the recording medium every time a writing command to the file system occurs. . As a result, the number of times of writing to the recording medium is reduced to improve the transfer speed.

  When the delayed writing mechanism is working, there is a timing when there exists data that exists only in the memory and is not written to the recording medium 610. If a failure occurs at that timing, the contents of the index file 612 and the video file 611 may be inconsistent. There are two types of mismatch cases: a case with few video files 611 (FIG. 12) and a case with few index files 612 (FIG. 13). FIG. 14 shows the configuration of a video restoration apparatus 1400 for dealing with these cases.

  In the video restoration apparatus 1400, the same components as those of the video recording apparatus 600 and the video restoration apparatus 1100 are assigned the same numbers, and the description thereof is omitted.

  The index reading unit 1402 reads the index file 612 and interprets the recorded index information. The index detection unit 1401 reads the video file 611 and creates index information that is not written in the index file 612. The index adjustment unit 1403 adds the index information detected by the index detection unit 1401 to the index file 612, or deletes unnecessary index information that does not exist in the video file 611 from the index file 612.

  The case where the video file 611 is small (FIG. 12) will be described with reference to the flowchart of FIG. In the index file reading process S151, the index reading unit 1402 reads the index file 612 and recognizes that there are 6 frames of data. In the data size calculation process S152, the index detection unit 1401 reads the video file 611 and calculates that the compressed data is 68,000 bytes. The index adjustment unit 1403 determines that 68,000 bytes from the index file 612 correspond to 4 frames of data in the unnecessary index determination S153, and therefore determines that there is an unnecessary index. Next, the index adjustment unit 1403 deletes unnecessary fifth and sixth frame index information from the index file 612 in the index deletion processing S156. In the index composition processing S79, the index composition unit 604 restores the video file 611 by combining the filler and the deleted index file 612 with the end of the video file 611 as described above.

  A case where the index file 612 is small (FIG. 13) will be described. In the index file reading process S151, the index reading unit 1402 reads the index file 612 and recognizes that there is 4-frame data. In the data size calculation process S152, the index detection unit 1401 reads the video file 611 and calculates that the compressed data is 82 KB. In the unnecessary index determination S153, the index adjustment unit 1403 determines that 82KB is data of 5 frames or more, and therefore determines that there is no unnecessary index. In the index detection process S154, the index detection unit 1401 inspects the compressed data of the fifth frame and thereafter, and extracts index information of the fifth and sixth frames (described later). The index adjustment unit 1403 adds the index information detected by the index detection unit 1401 to the index file 612, and restores the same index information as in FIG. In the index composition processing S79, the index composition unit 604 combines the filler and the added index file 612 at the end as described above.

  FIG. 16 shows the internal structure of an MPEG picture. In the picture, a picture header (Picture Header) is first arranged, and a slice (Slice) is arranged behind the picture header. The picture header includes a start code (Start Code), a temporal reference indicating the display order of images in the GOP, and a picture type (Picture Type).

  The index detection process S154 will be described with reference to the flowchart of FIG. The index detection unit 1401 executes index detection processing S154. In the picture header search processing S171, the search is started from the 65,000th byte which is the stream offset of the index information of the last frame written in the index file, and the start code is found at the 68,000th byte. In the picture header presence / absence determination S172, since the picture header is found in the picture header search process S171, it is determined that there is a picture header. In the index creation processing S173, the temporal reference of the picture in the fifth frame is 5 (0 reference), so that it is understood that it is output sixth in the GOP. Therefore, it can be seen that the Temporary Offset of the index information is a value 1 obtained by subtracting the frame number 5 in the recording order from the temporal reference +1 (0 is necessary because it is 0). Key-Frame Offset is -3 obtained by subtracting frame number 5 from frame number 2 in the I-picture display order. In the Stream Offset, 68,000 bytes of the start position of the picture header is set.

  Next, the picture header search process S171 starts the search from the 68,000th byte and finds the start code at the 79,000th byte. In the index information creation process S173, a value −1 obtained by subtracting the recording frame number 6 from the temporal reference +1 (5) is set in the Temporary Offset. Key-Frame Offset is set to -4 obtained by subtracting frame number 6 from frame number 2 in the display order of the I picture. In the Stream Offset, 79,000 bytes as the start position of the picture header are set. Since Flags is a picture type in display order, the picture type P of the sixth frame is set to the fifth index, and the picture type B of the fifth frame is set to the sixth index. In this way, the same index information as in FIG. 5C can be configured.

  By these methods, one correct video file 611 can be obtained even in the case where a mismatch between the index file 612 and the video file 611 occurs.

  The video recording apparatus according to the present embodiment can finish the recording end process in a short time. Further, when a failure occurs during the recording and the recording is not completed normally, the video recording apparatus relies on the index file. Since the repair process can be advanced, it can be applied to a commercial or consumer video camera or the like, which is useful.

100 Physical Drive 101 Logical Drive 102 Master Boot Record 103 Partition Boot Sector 104 File Allocation Table # 1
104b File allocation table # 2
105 root directory entry 106 user data area 107 directory entry 108 first cluster number 110 cluster 111 cluster number 600 video recording device 601 encoding unit 602 video writing unit 603 index recording unit 604 index composition unit 610 recording medium 611 video file 612 index file 1100 1400 Video restoration device 1101 Data restoration instruction unit 1401 Index detection unit 1402 Index reading unit 1403 Index adjustment unit

Claims (3)

An encoding unit that encodes an externally input video signal using an inter-frame compression method;
A video writing unit for recording the video data encoded by the encoding unit as a video file on a recording medium;
An index recording unit that creates index information of the video data encoded by the encoding unit and records the index information on the recording medium;
An index composition unit for compositing the video file and the index file into one after recording the video file and the index file on the recording medium;
A video recording apparatus comprising: An index composition unit for compositing a video file existing on a recording medium and an index file related to the video file into one;
A data recovery instruction unit that confirms the consistency between the video file and the index file, and instructs the index composition unit to combine the video file and the index file together when the consistency is achieved. When,
A video restoration device comprising: An index reading unit for reading index information from the index file;
An index detector for creating index information from the video file;
Check the consistency between the index information read from the index reading unit and the index information created by the index detection unit, and add an index to the index information of the index file if consistency is not achieved Or an index adjustment unit for performing deletion,
The video restoration device according to claim 2, further comprising:

Images