TWI412268B - Method of de-interlacing a horizontal moving subtitle - Google Patents

Abstract

The present invention is directed to a method of de-interlacing a horizontal moving subtitle. A field is divided into regions, one of which is to be judged. Each region is subjected to moving/serration detection in order to obtain the number of moving pixels/serration pixels. If the to-be-judged region has moving pixels/serration pixels larger than other regions, the to-be-judged region is then de-interlaced by motion adaptive de-interlace and other regions are de-interlaced by 3: 2/2: 2 film-mode de-interlace; otherwise, the whole field is de-interlaced by 3: 2/2: 2 film-mode de-interlace.

Description

Deinterlacing method for horizontally moving subtitles

The present invention relates to de-interlaces, and more particularly to a de-interlacing method for horizontally moving subtitles.

The National Television System Committee (NTSC) broadcast video format has an image frequency of 60 fields per second (ie, 60 fields/sec), whereas the film frequency of movies taken in film is 24 per second. Frame (ie, 24 frames/sec). In order to enable the movie to be played on an NTSC-formatted TV set, the first frame in the adjacent two frames of the movie is converted into a 3-picture field by a 3:2 pull down method. The second frame is converted to 2 fields; in other words, the adjacent frames are converted to 5 fields. According to this principle, 24 frames per second is converted into 60 fields per second, thus conforming to the NTSC format, which is called 3:2 film mode. formula.

The image frequency of the Phase Alternating Line (PAL) broadcast video format is 50 fields per second (ie, 50 fields/sec). In order to allow the movie to play on a PAL format TV, you can use a 2:2 pull down method to convert 24 frames per second to 48 fields per second, however this will result in about 4% accelerate. In view of this, you can use another improved 2:2 conversion, called 2:2:2:2:2:2:2:2:2:2:2:3 conversion, which can take 24 pictures per second of the movie. The box is converted to 50 fields per second, thus conforming to the PAL format, which is called the 2:2 film mode. In addition, when NTSC is in progressive mode, it converts 30 frames per second into 60 fields per second, which is also a 2:2 film mode.

NTSC or PAL format transmitters are interlaced field images. To display them on a non-interlaced or progressive display, you must first restore the field to a map. Box, this is called de-interlace. In general, interleaved signals use spatial interpolation or temporal interpolation to create or fill missing scan lines. Wherein spatial interpolation uses only pixels of the same field to generate new pixels, and is therefore also referred to as intra-field or two-dimensional interpolation; temporal interpolation uses adjacent different fields. Pixels are used to generate new pixels, and are therefore also referred to as inter-field or three-dimensional interpolation. Typically, the quiescent zone is deinterleaved by temporal interpolation, while the mobile zone is spatially interpolated for deinterlacing. The still area and the moving area must be discriminated by motion detection. Therefore, deinterlacing by interpolation is also called motion adaptive de-interlace method to distinguish it from the film mode. Film mode de-interlace method.

However, for the deinterlacing of the film mode signal, all the scanning lines of the previous scene or the latter field are copied to be reduced to a frame, which is different from the interleaving of the interlaced signals. Deinterlacing the film mode signal by interpolation will blur or beat the picture.

When receiving a TV signal, the video content that may be received is in the movie mode, but the horizontal mobile interlaced signal is added by the video player or the fourth system vendor. At this time, if the film mode is used for interlacing, the subtitle font will be broken; if the motion adjustment is used to interleave, the image will be blurred.

Therefore, there is a need to propose a novel method for improving the cracking of the horizontal moving font of the film mode.

In view of the above, one of the objects of the embodiments of the present invention is to provide a de-interlacing method for horizontally moving subtitles, which can be deinterleaved regionally for horizontally moving subtitles without jumping out of the movie mode.

According to an embodiment of the invention, a field is divided into multiple regions, wherein one region is a region to be determined, and the moving pixel or the sawtooth pixel of each region is counted by a motion detection or a sawtooth detection method. It is determined whether the number of moving pixel points or sawtooth image points of the to-be-determined area is greater than the average moving pixel point or the average sawtooth image number of each area. If the result of the determination is yes, then motion adaptive de-interlace is used for the region to be determined, and the remaining regions are used for film mode de-interlace by 3:2 or 2:2. If the result of the determination is no, the entire field is deinterleaved using a 3:2 or 2:2 film mode.

11-15A‧‧ (Step A) steps

11-15B‧‧ (Step B) steps

41-43A‧‧ (Step 4A) steps

41-43B‧‧ (Step 4 B) steps

The flowchart of FIG. 1A shows a deinterlacing method of a horizontal moving subtitle to be determined region according to an embodiment of the present invention.

The flowchart of the first B diagram shows a deinterleaving method of another to-be-determined area of the horizontal moving subtitle of the embodiment of the present invention.

The second figure shows that an entire field is divided into four equal areas.

The third figure shows a schematic diagram of using a motion detection method to obtain moving pixel points.

The flowchart of FIG. 4A shows a deinterleaving method of a horizontal moving subtitle to be determined region according to another embodiment of the present invention.

The flowchart of FIG. 4B shows a deinterleaving method of another to-be-determined area of the horizontal moving subtitle according to the embodiment of the present invention.

The fifth figure shows a schematic diagram using a sawtooth detection method to obtain sawtooth pixel points.

The flow chart of the first A picture and the first B picture shows a deinterlacing method of one of the horizontal moving subtitles to be judged in the embodiment of the present invention, in order to avoid the font rupture phenomenon of the horizontally moving interlaced subtitle in the 3:2 film mode. . Although the horizontally moving subtitle of the present embodiment is moved from right to left, it may be moved from left to right in other embodiments.

For the 3:2 film mode, the field is A1 A2 A3 B4 B5 C6 C7 C8 D9 D10. When using A1 and A3 for motion detection, since the two belong to the same frame, the number of moving pixels obtained when they are subtracted will be small; when A2 and B4 are used for motion detection, The number of moving pixels obtained will be relatively large. The traditional method is that when the screen appears to move the subtitle horizontally, it is often impossible to find the interlaced subtitles in time, and continue to use the 3:2 video mode. Deinterlace, which causes the font to rupture. This font rupture usually takes a long time to be judged.

In this embodiment, when A1 and A3 (that is, two fields having the same field picture (that is, the same parity field) and spaced by one field) are used for motion detection, an entire picture field is used ( That is, the field A2) is divided into four regions that are substantially equal: A region, B region, C region, and D region, which are located at the upper left, upper right, lower left, and lower right, respectively, as shown in the second figure. Although this embodiment divides the entire field into equal four regions, in other embodiments, it can be divided into other numbers, and the regions do not have to be equal. Furthermore, the entire map does not need to be divided. field.

Please refer to FIG. 1A. First, in step 11, the moving pixel points of each area (A area, B area, C area, and D area) are counted by the motion detection method, thereby obtaining the number of moving pixel points of each area. MA, MB, MC, and MD. The third figure shows a schematic diagram of using a motion detection method to obtain moving pixel points. In the illustration of the third figure, it is determined whether the image point c in the field A2 is a moving image point by calculating the absolute difference between the corresponding (position) pixels pa and na of the fields A1 and A3 (for example, gray scale Absolute difference), if it is greater than the preset threshold, the number of points ca is counted as a moving pixel. Another embodiment of the motion detection method opens a window in each of the fields A1 and A3 (for example, the window contains 3*3, 3*1, 1*3 image points), respectively, including pixels pa and na, and a calculation window. (for example, the field A1) and the sum of the absolute differences of the corresponding (position) images in the window of another field (for example, the field A3) (sum of absolution difference, SAD, for example, the absolute of the grayscale The sum of the differences), if it is greater than the preset threshold, the number of points ca is counted as a moving pixel.

Next, in step 12A, it is determined whether the number of moving pixel points of the D zone (to be determined) is greater than the average number of moving pixels of each zone. Usually, the judgment of step 12A has to be performed a plurality of times, for example, three times, and is controlled by the K value in the flow of the second figure. The judgment formula of this step 12A Can be expressed as follows: MA+MB+MC+MD<T3:2, and MD>T3:2/4

Among them, T3:2 is the default 3:2 film mode threshold.

If it is determined by step 12A that the number of moving pixel points in the D zone is greater than the average number of moving pixels in each zone (ie, (MA+MB+MC+MD)/4), it can be considered that the horizontally moving subtitle starts to appear in D. Area. At this time, in step 13A, motion adaptive de-interlace is used for the D zone, and the A, B, and C zones continue to use the 3:2 film mode de-interlace. Until the entire field does not meet the 3:2 film mode condition, the whole picture is switched to the mobile adjustment to interleave. If the decision of step 12A does not match, the entire field continues to be interleaved using the 3:2 film mode (step 14).

In the deinterlacing, since the contrast or color of the subtitle area is usually obvious, the sum of the absolute difference or the absolute difference (SAD) obtained after the motion detection is large. In view of this characteristic, in the present embodiment, the threshold value T can be dynamically increased (step 15A); when the absolute difference or the sum of absolute differences (SAD) is greater than the critical value, the corresponding pixel in the region to be determined (third map) , like the number of points ca) will be changed to two-dimensional de-interlace (spatial de-interlace), on the contrary, the corresponding pixel points in the area to be judged continue to use the 3:2 film mode to interleave.

Please refer to FIG. 1B, which shows a flowchart of a deinterlacing method for another to-be-determined area of the horizontal moving subtitle according to the embodiment of the present invention. Actually, since it is impossible to determine in advance the area in which the horizontal moving subtitle appears, if it is determined that the horizontal moving subtitle is not in the D area, the determination is made for the B area. In the same principle, it is determined in step 12B whether the number of moving pixel points in the B area (to be determined) is greater than the number of moving pixels in each area. Usually, the judgment of step 12B has to be performed multiple times. For example, 3 times, in the flow of the second figure, it is controlled by the K value. The judgment formula of this step 12B can be expressed as follows: MA+MB+MC+MD<T3:2, and MB>T3:2/4.

If it is judged by step 12B that the number of moving pixel points of the B area is larger than the average number of moving pixel points (ie, (MA+MB+MC+MD)/4), it can be considered that the horizontal moving subtitle starts to appear in the B area. At this time, in step 13B, the mobile region is deinterleaved for the B region, and the A, C, and D regions continue to use the 3:2 film mode to interleave until the entire field does not meet the 3:2 film mode condition. , then switch to the mobile adaptation to interlace. If the decision of step 12B does not match, the entire field continues to be interleaved using the 3:2 film mode (step 14).

The flow charts of the fourth A picture and the fourth B picture show a deinterlacing method of one of the horizontal moving subtitles to be judged in another embodiment of the present invention, to avoid horizontally moving the subtitle font in the 2:2 film mode. Cracking phenomenon. Although the horizontally moving subtitle of the present embodiment is moved from right to left, it may be moved from left to right in other embodiments.

For the 2:2 film mode, the field is A1 A2 B3 B4 C5 C6 D7 D8. When using A1 and A2 for jagged detection, since the two belong to the same frame, the number of moving pixels obtained will be small; when using A2 and B3 for sawtooth detection (not belonging to the same frame), The resulting number of jagged pixels will be relatively large. The traditional method often fails to find interlaced subtitles in time when the screen appears to move the subtitles horizontally, and continues to deinterlace in the 2:2 film mode, thus causing the font to burst. This font rupture usually needs to wait After a long time, it will be judged.

In this embodiment, when the A1 and A2 (that is, the adjacent two fields) are saw-toothed, an entire field (ie, the field A1) is divided into four regions that are substantially equal: A Zones, Zone B, Zone C, and Zone D are located in the upper left, upper right, lower left, and lower right, respectively, as shown in the second figure. Although this embodiment divides the entire field into equal four regions, in other embodiments, it can be divided into other numbers, and the regions do not have to be equal. Furthermore, the entire map does not need to be divided. field.

Please refer to FIG. 4A. First, in step 41, the zigzag pixel points of each area (A area, B area, C area, and D area) are counted by a sawtooth detection method, thereby obtaining the number of zigzag pixels in each area. For FA, FB, FC and FD. The fifth figure shows a schematic diagram using a sawtooth detection method to obtain sawtooth pixel points. In the illustration of the fifth figure, it is determined whether the image point pb in the field A1 is a sawtooth image point. In this embodiment, three differences are obtained for the field A1 and the field A2: the corresponding phase of the field A1 is obtained. The adjacent pixels pa and pc are subtracted, and the corresponding adjacent pixels pa and cb of the field A1 and the field A2 are subtracted, and the corresponding adjacent pixels pc and cb of the field A1 and the field A2 are subtracted. If the difference of the same field A1 is smaller than the difference between the adjacent fields A1, A2, the image point pb is counted as a sawtooth pixel.

Next, in step 42A, it is determined whether the number of zigzag pixels in the D zone (to be determined) is greater than the average number of zigzag pixels in each zone. Usually, the judgment of step 42A has to be performed a plurality of times, for example, five times, and is controlled by the K value in the flow of the fourth A diagram. In general, since the judgment of the 2:2 film mode is usually difficult to judge in the 3:2 film mode, the number of judgments of the 2:2 film mode is generally greater than the number of judgments of the 3:2 film mode. The judgment formula of this step 42A can be expressed as follows: FA+FB+FC+FD<T2:2, and FD>T2: 2/4

Among them, T2:2 is the default 2:2 film mode threshold.

If it is judged by step 42A that the number of zigzag pixels in the D zone is larger than the average number of zigzag pixels in each zone (ie, (FA+FB+FC+FD)/4), it can be considered that the horizontally moving subtitle starts to appear in D. Area. At this time, in step 43A, motion adaptive de-interlace is used for the D zone, and the A, B, and C zones continue to use the 2:2 film mode de-interlace. Until the entire field does not meet the 2:2 film mode condition, the whole picture is switched to the mobile adjustment to interleave. If the decision of step 42A does not match, the entire field continues to be interleaved using the 2:2 film mode (step 44). Wherein, when the D zone is deinterleaved using the motion adaptation, the two-dimensional deinterlacing can be performed only for the number of sawtooth images in the B zone.

Please refer to FIG. 4B, which shows a flowchart of a deinterleaving method for another to-be-determined area of the horizontal moving subtitle according to the embodiment of the present invention. Actually, since it is impossible to determine in advance the area in which the horizontal moving subtitle appears, if it is determined that the horizontal moving subtitle is not in the D area, the determination is made for the B area. In the same principle, it is determined in step 42B whether the number of zigzag pixels in the B region (to be determined) is greater than the average number of zigzag pixels in each region. Generally, the judgment of step 42A has to be performed a plurality of times, for example, five times, and is controlled by the K value in the flow of the fourth diagram B. The judgment formula of this step 42B can be expressed as follows: FA + FB + FC + FD < T2: 2, and FB > T2: 2 / 4.

If it is judged by step 42B that the number of zigzag pixels in the B region is larger than the average number of zigzag pixels in each region (ie, (FA+FB+FC+FD)/4), it can be considered that the horizontally moving subtitle starts to appear in B. Area. At this time, in step 43B, the mobile adaptation is deinterleaved for the B zone, and the A, C, and D zones continue to use the 2:2 film mode to interleave until the entire field does not meet the 2:2 film mode condition. , then switch to the mobile adaptation to interlace. If the decision of step 42B does not match, the entire field continues to be interleaved using the 2:2 film mode (step 44). Wherein, when the B area is moved and deinterleaved by using the motion adjustment, the two-dimensional deinterlacing can be performed only for the number of sawtooth images in the B area.

The above description is only the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention; all other equivalent changes or modifications which are not departing from the spirit of the invention should be included in the following Within the scope of the patent application.

11-15A‧‧‧Steps

Claims (7)

A de-interlacing method for horizontally moving subtitles includes: dividing a field into multiple regions, wherein one region is a region to be determined; counting moving pixel points of each region by using a motion detection method; and determining a moving pixel of the region to be determined Whether the number is greater than the average number of moving pixels of each zone; if the result of the determination is yes, the motion adaptive de-interlace is used for the to-be-determined zone, and the remaining zones use the 3:2 film mode. Film mode de-interlace; and if the result of the determination is no, the entire field is deinterleaved using a 3:2 film mode. The deinterlacing method for horizontally moving subtitles according to claim 1, wherein the motion detection method has two fields of the same field and spaced apart from the field, and the absolute difference of the corresponding pixels is greater than one. When the threshold is preset, the corresponding image of the field is counted as a moving pixel. For example, in the deinterlacing method of the horizontal moving subtitle described in claim 2, when the determination result is yes, the method further includes dynamically increasing the preset threshold, so that when the absolute difference is greater than the preset threshold, the The corresponding pixels in the area to be judged will be deinterlaced in two dimensions. The deinterlacing method for horizontally moving subtitles according to claim 1, wherein the motion detection method has two fields with the same field picture and spacing the picture field, and each of the two fields is opened. The window calculates a sum of absolute differences (SAD) between the respective pixels in the window and the corresponding image in the other field window. If it is greater than a predetermined threshold, the corresponding image of the field is counted as a moving pixel. For the deinterlacing method of the horizontal moving subtitles in claim 4, when the above judgment result is yes, the dynamic inclusion of the preset threshold is further included, so that the absolute difference is When (SAD) is greater than the preset threshold, the corresponding pixels in the to-be-determined area will be de-interlaced in two dimensions. The deinterlacing method for horizontal moving subtitles according to claim 1, wherein the field is divided into A area, B area, C area and D area, respectively located at upper left, upper right, lower left and lower right and from A One of the district, B zone, C zone and D zone is the zone to be judged. The deinterlacing method for horizontally moving subtitles according to item 6 of the patent application scope, wherein the determining step uses the following judgment formula: MA+MB+MC+MD<T _3:2 , and MX>T _3:2 /4 Among them, MX, MA, MB, MC, and MD are the number of moving pixels in the to-be-determined area, A-area, B-area, C-area, and D-area, respectively, and T _3:2 is the preset 3:2 film mode criticality. value.