CN102655582B - Apparatus and method for detection of moving video aspect - Google Patents

Abstract

A device and method for detecting dynamic video sample. The apparatus for detecting dynamic video patterns comprises a motion detection unit, a cyclic detection unit, a video detection unit and a de-interlacing unit. The motion detection unit receives a plurality of continuous scene images for analyzing and determining whether each scene image has motion. A first value is given if it belongs to a scene that is not moving and a second value is given if it belongs to a scene that is moving, so as to obtain a string of values. The loop detection unit receives the string of values and detects whether the first value or the second value has a loop period to output a loop determination value. The film detection unit receives the string of values and the cycle decision value to determine whether the received scene images belong to a film format, and outputs a change state. And the de-interlacing unit restores the scene images into an original film according to the change state of the film detection unit.

Description

Apparatus and method for detection of moving video aspect

technical field

The invention relates to a technology for detecting video cadences, in particular to a detection technology for moving video cadences.

Background technique

Because the shooting video rate is generally different from the broadcasting standard. The video recording rate is, for example, 24Hz. Common broadcasting standards are, for example, 60 Hz for NTSC and 50 Hz for PAL & SECAM. Therefore, in order to make the filmed video playable in standards such as NTSC or PAL & SECAM, the filmed film will be pulled down to the desired broadcasting standard.

For example, when changing from 24Hz to PAL standard 50Hz, one captured image will be disassembled into two, including even and odd images. As is generally known, an even image is formed by extracting images belonging to even scan lines from the captured images, and an odd image is formed by extracting images belonging to odd scan lines from the captured images. This changes the 24Hz to a wave square rate of about 50Hz. In other words, the captured original image will be disassembled according to the selected video format at the rate required by the playback standard, and played in an odd/even interlaced manner.

Conversely, if the played field image is to be restored to the original captured image, then it needs to detect its video mode, and restore the original captured image according to the video mode.

FIG. 1 shows a schematic block diagram of a conventional system for restoring a played scene image to an original film. Referring to FIG. 1 , the motion detection unit (motion detection) 100 needs to refer to previous scene images and subsequent scene images to determine whether the current scene image is moving for the current scene image to be processed. For example, if the current scene image and subsequent scene images are even images and odd images from the same captured image, a result of no movement will be obtained. If the current scene image and subsequent scene images are even and odd images from different captured images, a moving result will be obtained. In other words, each scene image will be detected for motion to obtain a style.

Identify the change pattern unit (cadence pattern recognition) 102, compare the pattern detected by the movement detection unit 100 with the data in the pattern data unit (cadence database) 104, and obtain the corresponding change pattern (cadence pattern), Finally, the de-interlacing unit 106 is controlled according to the changing state to restore each captured image, which is also called a frame (frame).

In the traditional method mentioned above, to correctly restore the scene image to the original image requires the aspect data unit 104 to store various changes. In addition to using the aspect data unit 104 to store various variations, if more types of variations are used without being stored in the aspect data unit 104 , it cannot operate correctly.

Contents of the invention

The present invention provides a movable video pattern detection device and method, which can detect what kind of changing patterns a film has, without being limited to a fixed number of changing pattern detections.

An embodiment of the present invention provides a moving video pattern detection device, which includes a motion detection unit, a cycle detection unit, a video detection unit and a deinterlacing unit. The motion detection unit receives a plurality of consecutive scene images for analyzing and determining whether each scene image is moving. If it belongs to a non-moving scene, a first value is given, and if it belongs to a moving scene, a second value is given to obtain a series of values. The cycle detection unit receives the string of values and detects whether the first value or the second value has a cycle period, so as to output a cycle decision value. The film detection unit receives the string of values and the cycle decision value to determine whether the received scene images belong to the film format, and outputs a change pattern. The de-interlacing unit restores the scene images into an original video according to the change of the video detection unit.

An embodiment of the present invention also provides a detection device for detecting a film and an interlaced pattern, including a motion detection unit, a loop detection unit, and a film detection unit. The motion detection unit receives a plurality of consecutive scene images for analyzing and determining whether each scene image is moving. If it belongs to a non-moving scene, a first value is given, and if it belongs to a moving scene, a second value is given to obtain a series of values. The cycle detection unit receives the string of values to detect whether the first value or the second value has a cycle, and outputs the cycle. The video detection unit receives the series of values and determines whether it is a video format according to the cycle period, and judges an interlaced pattern according to the change of the first value and the second value within the cycle period.

An embodiment of the present invention also provides a method for detecting a film and an interlaced pattern, which is used for detecting a plurality of scene images successively input to a detection device. The method includes analyzing and determining whether each scene image is moving, and giving a first value if it belongs to a non-moving scene, and giving a second value if it belongs to a moving scene, so as to obtain a series of values. In addition, the method further includes detecting whether the string of values has a cycle period of the first value or the second value. Determine whether it belongs to a video format according to the string of values. A staggered pattern is determined by the change of the first value and the second value within the cycle period.

In order to make the above-mentioned features and advantages of the present invention more comprehensible, the following specific embodiments are described in detail with reference to the accompanying drawings.

Description of drawings

FIG. 1 is a schematic block diagram of a traditional system for restoring a played scene image to an original film.

FIG. 2 shows a schematic diagram of a disassembly mechanism using 2:2 according to an embodiment.

FIG. 3 shows a schematic diagram of a disassembly mechanism using 2:3 according to an embodiment.

FIG. 4 shows a schematic diagram of the relationship between the numerical string and the change pattern in FIG. 3 .

FIG. 5 shows a schematic diagram of analyzing numerical strings according to a predetermined number.

FIG. 6 shows a schematic structural diagram of a movable video aspect detection device according to an embodiment.

Description of reference symbols

100: mobile detection unit

102: Identifying Variation Aspect Units

104: Aspect data unit

106: de-interleaving unit

200: mobile detection unit

202: Loop detection unit

204: Film detection unit

206: de-interleaving unit

Detailed ways

The present invention can directly detect changes from the played scene images without being limited to a fixed number of changes, and thus can respond to various changes. The following examples are given to illustrate the present invention, but the present invention is not limited to the examples given. In addition, the possibility of appropriate combination is also allowed among the cited embodiments.

The mechanism for detecting whether there is movement in each scene image is described below.

Take the 3:2 disassembly as an example, which means that the previous image will be disassembled into three scene images, and the latter image will be disassembled into two scene images. After being detected by the motion detection unit, a sequence of 10100 patterns will be obtained from the latest scene image to the oldest scene image, wherein 1 represents movement, and 0 represents no movement.

FIG. 2 shows a schematic diagram of a disassembly mechanism using 2:2 according to an embodiment. Referring to FIG. 2 , the captured video is taken as a frame (picture frame), and its shooting rate is, for example, Hz. In this embodiment, four frames (frames) are taken as an example, such as frame (frame) 1-frame (frame) 4 . Since this embodiment is a 2:2 disassembly mechanism, each frame will be sequentially disassembled into an even image (Even) and an odd image (Odd) up to 50/60Hz scene images. This embodiment starts with an even image, but it is also possible to start playing with an odd image. After the frames are disassembled into even images and odd images, the even images and odd images are combined into scene images according to broadcasting standards such as playing on TV. The scene images generated by frame 1-frame 4 are, for example, 1-8. Taking frame 1 as an example, since scene image 1 and scene image 2 come from the same frame 1, the images are weaved from even images and odd images, so there is no difference between scene image 1 and scene image 2 move, so it is marked as 0. Then, the scene image 2 is compared with the scene image 3, and it is found that there is movement. This is because scene image 2 is the image of frame 1, and scene image 3 is the image of frame 2, unless it is a static image, scene image 2 will always be marked with 1.

By analogy, other scene images 3-8 will also have the same cycle phenomenon, but have specific changes. This example is 01010101. Since each scene image can only have a state of 0 or 1, one bit can be used to represent a binary value to form a value string. Here, 0 and 1 are binary values, but in general, different values can be used to mark the two states, not limited to the way of marking 0 and 1. In other words, a situation with no movement can be represented by one value, and a situation with movement can be represented by another value.

With respect to the embodiment in FIG. 2 , in the value string, for example, the value 0 of the first bit starts, and it repeats every two bits, as shown by the arrow.

FIG. 3 shows a schematic diagram of a disassembly mechanism using 2:3 according to an embodiment. Refer to Figure 3. Due to the 2:3 disassembly mechanism, frame 2 will be disassembled into three images. According to the same mechanism described in Figure 2 above, the scene images adjacent to the new frame will show movement. And marked as 1. In this embodiment, the variation is the repetition of 01001.

FIG. 4 shows a schematic diagram of the relationship between the numerical string and the change pattern in FIG. 3 . Take the 20-digit value string for analysis. If the detection cycle period is 5, for example, the value 0 of the first bit on the left is used for inspection, and it is expected that the cycle will repeat every 5 bits. If the check does not correspond to a cycle of 5, the value 0 will not cycle, but will not represent the correct cycle.

In addition, each value 1 is preceded by 0, which means that there is a frame change, which belongs to the dynamic display of the movie. If only the result conforms to the fact that each value 1 is followed by 0 and has a fixed cycle period, it can be inferred that it is a movie, and the original movie frame can be restored from the changed state of the cycle period.

FIG. 5 shows a schematic diagram of analyzing numerical strings according to a predetermined number. Referring to FIG. 5 , in order to effectively detect various possible cycle periods or changes, it can be set to take a predetermined number of value strings. In this embodiment, for example, a numerical string of 31 digits is used for analysis. After obtaining the numerical string corresponding to whether there is movement, detection may be performed on several possible cycle periods or optional cycle periods to check whether there is a cycle period. The cycle times of inspection are for example 5, 10, 12, 15, 25 etc., but these are not the only ways. The changes included in cycle 5, 10, 12, 15, and 25 correspond to different dismantling methods:

Cycle 5: (32);

Cycle 10: (22), (2224), (2332), (55), (64);

Loop cycle 12: (32322);

Loop cycle 15:(87);

Cycle 25: (222222222223).

As far as the actual cycle period is selected, the cycle period n may be a value selected from a number of factor values in 50/60 Hz.

The disassembly method in FIG. 5 is, for example, 3:3:2:2, which means that the first frame in the original frame is disassembled into three images mixed with even images and odd images. The second frame is disassembled into three images like the first frame. The third frame is disassembled into two images. The fourth frame is also disassembled into two images. In addition, the disassembled value string of 2:3:3:2 and the disassembled value string of 3:3:2:2 also have a similar circular relationship, which is 01001001010100100101. The disassembled images maintain the order of even-odd or even-odd. In this embodiment, the value 0 of the first bit is taken as a reference. If the case of cycle period 5 is checked, the sixth bit is 1, so there is no case of cycle period 5. If the case of a cycle period of 10 is examined, the value of every 10th bit repeats to be 0, so there is a case of a cycle period of 10. If we examine the case of cycle 12, the 13th bit is 1, so there is no case of cycle 12. If we examine the case of cycle 15, the 16th bit is 1, so there is no case of cycle 15. If the case with a cycle time of 25 is checked, the 26th bit is 1, so there is no case with a cycle time of 25. Also, if the case of cycle 5 is examined, it is a factor of cycle 10, so it can still be attributed to the case of cycle 10. In this way, the change pattern can be found out. The cycle 10 includes many possible dismantling methods, but according to the change pattern of the cycle cycle 10, the combination method can still be correctly calculated to restore the frame data.

This value string is the playback format of the corresponding video. In this embodiment, under the detection mechanism for determining whether the scene image is a dynamic movie, the values before and after each value 1 are 0.

Based on the above mechanism and method, it can be implemented by a movable video aspect detection device. FIG. 6 shows a schematic structural diagram of a movable video aspect detection device according to an embodiment. The moving video pattern detection device includes a motion detection unit 200 , a cycle detection unit 202 , a film detection unit 204 and a deinterlacing unit 206 . The motion detection unit 200 receives a plurality of consecutive scene images for analyzing and determining whether each scene image has motion. If it belongs to a non-moving scene, a first value, such as 0, is given. If the scene image belongs to a moving scene, a second value such as 1 is given, so as to obtain a series of values, such as a bit string value. The cycle detection unit 202 receives the string of values and detects whether the first value or the second value has a cycle period, so as to output a cycle decision value. The cyclic decision value is any suitable form of decision value that can indicate the existence of a cyclic period, and is not limited to a specific method or information.

Next, the video detection unit 204 receives the string of values and the loop determination value to determine whether the received scene images belong to the video format, and outputs a change pattern. The de-interlacing unit 206 restores the scene images into an original video according to the changes of the video detection unit. The de-interlacing unit 206 can learn how to decompose the original frame into a scene image according to the change pattern, and can substantially restore the data of the original frame.

In addition, when the motion detection unit 200 and the deinterleaving unit 206 are performing their functions, in addition to the current scene image to be processed, they also refer to the adjacent previous scene image or subsequent scene image or both, It depends on the mechanism of motion detection and de-interleaving, and is not limited to a specific method. Several embodiments of the present invention are to generate a numerical string, and directly detect changes from the numerical string, so as to respond to various moving video states.

Although the present invention has been disclosed above with the embodiments, it is not intended to limit the present invention. Those skilled in the art can make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, the present invention The scope of protection is based on the claims of the present invention.

Claims (20)

1. A movable video pattern detection device, comprising: A motion detection unit, which receives a plurality of consecutive scene images, and analyzes and determines whether each scene image is moving. If it belongs to a non-moving scene, it will give a first value; if it belongs to a moving scene, it will give a first value. Binary value to get a string of values; A cycle detection unit, in the first stage, receives the string of values, and only based on a set of candidate cycles, detects whether the first value or the second value has a cycle, so as to output a cycle decision value; A film detection unit, in the second stage following the first stage, receives the series of values and the loop decision value to determine whether the received scene images belong to the movie format, and outputs a change pattern, wherein the loop decision A value corresponds only to a set of candidate variants for which the variant is one of the set of candidate variants; and A de-interlacing unit restores the scene images into an original video according to the change of the video detection unit. 2. The movable video pattern detection device as claimed in claim 1, wherein the motion detection unit compares a current scene image with an adjacent subsequent scene image or a previous scene image to determine whether the current scene The image has moved. 3. The movable video pattern detection device as claimed in claim 2, wherein the first value and the second value are recorded by one bit, forming a predetermined number of a series of bits. 4. The movable video pattern detection device as claimed in claim 3, wherein the loop detection unit is to detect whether a binary value for a selected bit is repeated at intervals of a detection cycle, and the consecutive multiple strings bits are the same detection period, wherein the determined detection period is the cycle period. 5. The movable video aspect detection device as claimed in claim 4, wherein the cycle period comprises a period of checking every n bits, wherein n is a value selected from a plurality of factor values in 50/60 Hz . 6. The movable video pattern detection device as claimed in claim 4, wherein the film detection unit obtains the cycle period, and simultaneously detects the two before and after the bit representing the moving second value in the string The bit is the first value that represents no movement, to determine whether it belongs to the video format. 7. The movable video pattern detection device as claimed in claim 4, wherein the scene images are composed of a plurality of even images and a plurality of odd images in an interlaced order, wherein for each original video, The image data of even scanning lines is used to form the corresponding even image, and the image data of odd scanning lines is used to form the corresponding odd image. 8. The mobile video aspect detection device as claimed in claim 4, wherein the first value is 0, and the second value is 1. 9. A detection device for detecting a film and an interlaced pattern, comprising: A motion detection unit, which receives a plurality of consecutive scene images, and analyzes and determines whether each scene image is moving. If it belongs to a non-moving scene, it will give a first value; if it belongs to a moving scene, it will give a first value. Binary value to get a string of values; A cycle detection unit, in the first stage, receives the string of values, and only based on a set of candidate cycles, detects whether the first value or the second value has a cycle, and outputs the cycle; and A film detection unit, in the second stage following the first stage, receives the series of values and determines whether it is a film format according to the cycle period, and by the change of the first value and the second value in the cycle period An interleaving pattern is determined, wherein the cycle decision value only corresponds to a group of candidate changing patterns, and the interleaving pattern is one of the group of candidate changing patterns. 10. The detection device for detecting video and interlaced patterns as claimed in claim 9, wherein the motion detection unit compares a current scene image with an adjacent subsequent scene image or a previous scene image to determine whether the The current scene image has moved. 11. The detection device for detecting film and interlace pattern as claimed in claim 9, wherein the first value and the second value are recorded by one bit, forming a predetermined number of a series of bits. 12. The detection device for detecting film and interlaced pattern as claimed in claim 11, wherein the loop detection unit is to detect whether a binary value for a selected bit is repeated at intervals of a detection cycle, and the consecutive multiple The serial bits are the same as the detection period, wherein the determined detection period is the cycle period. 13. The detection device for detecting film and interlaced pattern as claimed in claim 12, wherein the cycle period includes checking the detection cycle every n bits, where n is determined by a plurality of factor values in 50/60Hz selected value. 14. The detection device for detecting film and interlaced pattern as claimed in claim 12, wherein the film detection unit obtains the cycle period, and simultaneously detects two before and after the bit representing the moving second value in the string The bit is the first value representing no movement. 15. The detection device for detecting a film and an interlaced pattern as claimed in claim 12, wherein the scene images are composed of a plurality of even images and a plurality of odd images according to a sequence of an interlaced pattern. 16. A method for detecting film and interlaced patterns, for detecting a plurality of scene images successively input to the detection device, comprising: Analyze and determine whether each scene image is moving, if it belongs to a non-moving scene, give a first value, if it belongs to a moving scene, give a second value, to obtain a series of values; Detecting in the first stage, and determining whether the string of values has a cycle of the first value or the second value based only on a set of candidate cycles; Determine whether it belongs to a video format according to the string of values; and In the second stage following the first stage, an interleaving pattern is determined from the change of the first value and the second value within the cycle period, wherein the cycle decision value only corresponds to a set of candidate change patterns , and the interleaving pattern is one of the set of candidate changing patterns. 17. The method for detecting a movie and an interlaced pattern as claimed in claim 16, wherein the scene images are composed of a plurality of even images and a plurality of odd images according to a sequence of an interlaced pattern. 18. The method for detecting film and interlaced patterns as claimed in claim 17, wherein a current scene image is compared with an adjacent subsequent scene image or a previous scene image to determine whether the current scene image has movement. 19. The method for detecting film and interlaced aspect as claimed in claim 18, wherein the first value and the second value are recorded by one bit, forming a predetermined number of a series of bits. 20. The method for detecting film and interlaced pattern as claimed in claim 19, wherein the detection of the cycle period comprises: select a bit to be detected, and refer to a binary value of the bit; and Detecting whether the bit value appears repeatedly at intervals of a detection period, and the consecutive multiple strings of bits have the same detection period, wherein the determined detection period is the cycle period.