TW201328316A - Method for auto-detect image format and playing method applying the same - Google Patents

Abstract

An single-frame image format auto-detection method includes: dividing an single-frame image into a plurality of macro-blocks; dividing each of the macro-blocks into a plurality of sub-blocks; generating a meta-block in each of the sub-blocks; calculating a respective pixel luminance sum characteristic value of each of the meta-blocks; based on the pixel luminance sum characteristic values, calculating a first confidence between a left half and a right half of the single-frame image; based on the pixel luminance sum characteristic values, calculating a second confidence between a top half and a bottom half of the single-frame image; and based on the pixel luminance sum characteristic values, the first confidence and the second confidence, identifying a format of the single-frame image.

Description

Method for automatically detecting image format and playing method thereof

This case is about a method for automatically detecting an image/movie format and a method for playing the same.

Currently, three-dimensional (3D) displays (such as televisions, computer screens, etc.) have become popular. 3D displays can not only display 3D content, but also display 2D content. However, 3D displays use different playback modes for different content. At present, the single frame 3D format is divided into left and right side by side (SBS, Side-by-Side) and side by side (TB, Top-and-Bottom). Therefore, the display needs to distinguish three modes in which the image belongs to 2D content, 3D format SBS, and 3D format TB.

In the SBS image, the pictures of the left and right eyes are arranged side by side. The graphs of the left and right eyes are half resolution in the horizontal direction, so the sum of the left and right eyes is equal to the normal resolution. For example, if the resolution is 1920x1080, the resolution of the left and right eyes in the SBS 3D image is 960x1080.

In the TB 3D image, the vertical resolution of the left and right eyes is only half of the normal resolution. For example, if the resolution is 1920x1080, the resolution of the left and right eyes is 1920x540.

When playing 3D images/movies, the 3D display should switch its playback mode to 3D playback mode. Otherwise, the 3D movie can only be played as a 2D movie, and its playback quality is very poor. For example, when playing a 3D movie of SBS, if it fails to switch to 3D playback mode correctly, the viewer may see that the movie is divided into the left half and the right half.

Therefore, the present invention proposes a method for automatically detecting an image format and a method for playing the same, which can automatically detect a single frame image format to switch to the correct playback mode.

The disclosure relates to a method for automatically detecting an image/movie format, which detects whether the image format is SBS by detecting the similarity between the left half and the right half of the single frame image.

The disclosure relates to a method for automatically detecting an image/movie format, which detects whether the image format is TB by detecting the similarity between the upper and lower sides of a single frame image.

The disclosure relates to a method for automatically detecting an image/movie format, which is determined not to be in SBS format or TB format for a single frame image, and is determined to be a 2D format.

According to an exemplary embodiment of the present disclosure, an automatic single-frame image format detection method is provided, including: dividing a single-frame image into a plurality of macroblocks; and dividing each of the giant blocks into a plurality of sub-blocks; Arranging an original block in the sub-blocks; calculating a sum of luminance values of the pixels of each of the original blocks; and calculating a left half and a right half of the single-frame image according to the sum of the luminance values of the pixels a first trust degree; calculating a second trust degree between one of the upper half and the lower half of the single frame image according to the sum of the pixel brightness sum; and summing the feature value according to the pixel brightness sum The first trust degree of the frame image and the second trust degree, detecting one format of the single frame image.

According to another exemplary embodiment of the present disclosure, a single-frame video playback method is provided, including: dividing a single-frame image into a plurality of macroblocks; and dividing each of the macroblocks into a plurality of sub-blocks; Setting a primitive block in the sub-block; calculating a sum of luminance values of the pixels of each of the original blocks; determining, according to the sum of the luminance values of the pixels, whether the first half of the single-frame image is similar to a first The second half, if yes, determines that the single frame image is in a three-dimensional format; and plays the single frame image according to the determined image format.

In order to better understand the above and other aspects of the present invention, the following specific embodiments, together with the drawings, are described in detail below:

Because of the depth of field, the left and right sides of the SBS format image are similar, but not identical. Similarly, because of the depth of field, the top and bottom half of the TB format image are similar, but not identical. Therefore, in the embodiment of the present invention, the image is determined to be a 2D format or a 3D format according to the principle, and the 3D image format is determined to be an SBS format or a TB format according to the principle.

Referring now to FIG. 1, a flow chart of a method for automatically detecting an image format according to an embodiment of the present invention is shown.

In step 110, the entire single frame image is segmented into a plurality of macro-blocks. As shown in Figure 2, the entire image is divided into N _Marco chunks. For convenience of explanation, the numbers of the giant blocks 1, 2, ..., N _Marco . Furthermore, for example, the entire image is horizontally split into Part and vertically split into Part. .

Further, in this embodiment, the entire image is horizontally divided into Part and vertical division into Part. However, it is not limited to this case. In other possible embodiments of the present case, the entire image can be horizontally unevenly segmented. In other possible embodiments of the present invention, the entire image can be vertically unevenly segmented.

Further, in other possible embodiments of the present invention, the number of horizontal divisions of the entire image is not necessarily a power of two. In other possible embodiments of the present invention, the number of vertical divisions of the entire image is not necessarily a power of two. This is within the spirit of the case. However, if the number of partitions is a power of 2, it is convenient to address the memory address.

In step 120, each chunk is divided into a plurality of sub-blocks. As shown in Fig. 2, each chunk is divided into N _Sub subblocks. For convenience of explanation, the number of each sub-block 1, 2, ..., N _Sub . Going a step further, a giant block is split horizontally into Part and vertically split into Part. .

Further, in the present embodiment, the average level of one block is divided into Part and vertical division into Part. However, it is not limited to this case. In other possible embodiments of the present case, a large block can be horizontally unevenly divided. In other possible embodiments of the present invention, a giant block may be vertically unevenly segmented.

Further, in other possible embodiments of the present invention, the number of horizontal splits for one macroblock is not necessarily a power of two. In other possible embodiments of the present case, the number of vertical divisions of one giant block is not necessarily a power of two. This is within the spirit of the case.

In step 130, a meta-block is set in each sub-block. In this embodiment, the upper left corner of the original block is aligned with the upper left corner of the sub-block. The width W_ME of the original block is between 1 and the sub-block width W_SB; and the height H_ME of the original block is between 1 and the sub-block height H_SB. The width and height of the original block are not necessarily the power of 2 and can be designed by the designer.

If the second image is taken as an example, the entire image 200 is divided into 16 ( N _Marco = 16) giant blocks 210, wherein the image 200 is horizontally divided into 4 ( N _mac-H = 2) portions and is vertically Split into 4 ( N _mac-V = 2) parts. A macroblock 210 is divided into 32 ( N _Sub = 32) sub-blocks 220, wherein the macroblock 210 is horizontally divided into 2 ( N _sub-H =1) parts; the macroblock is vertically divided into 16 ( N _{sub -V} = 4) parts. The original block 230 is disposed within the sub-block 220.

Please refer to Figure 2 again. In the embodiment of the present invention, the address of the upper left pixel of the first macroblock is addressed as (x, y) = (0, 0). The address of the upper left corner of the second block is addressed as (x, y) = (W_MB, 0), where W_MB represents the width of the block (in pixels). The address of the upper left corner of the fifth block is addressed as (x, y) = (H_MB, 0), where H_MB represents the height of the block (in pixels). The address of the upper left corner of other giant blocks can be deduced by analogy.

In step 140, the projection sum (SOP, Sum of Projection) of each original block is calculated. Here, the definition of projection and SOP is as follows:

Where Luma represents the pixel brightness, i is the label of the giant block; and j is the label of the sub-block within the i-th block. That is, SOP _{i , j} represents the sum of all pixel luminances in the jth sub-block within the i-th macroblock.

In step 150, the SBS confidence and the TB confidence of the single frame image are calculated according to the SOP of each original block. Roughly speaking, SBS trust represents the similarity between the left and right halves of the image: and TB trust represents the similarity between the upper and lower half of the image.

In detail, the algorithm of SBS trust degree is as follows:

Among them, in the above formula, abs represents the absolute value. Fig. 3A shows a schematic diagram of the SBS trust degree according to the present embodiment. In Figure 3A, the two ends of the arrow represent the subtraction of the SOPs of the two original blocks.

Take Figure 2 as an example:

Diff _ SBS ₁ = abs (SOP1-SOP3) + abs (SOP2-SOP4)

Diff _ SBS ₂ = abs (SOP5-SOP7) + abs (SOP6-SOP8).

Diff _ SBS ₃ = abs (SOP9-SOP11) + abs (SOP10-SOP12).

Diff _ SBS ₄ = abs (SOP13-SOP15) + abs (SOP14-SOP16).

Among them, SOP1 represents the sum of the SOPs of all the original blocks of the first giant block, and the rest can be analogized. Therefore, in the present embodiment, the SBS trust degree is the sum of the pixel brightness of one of the original blocks in the left half of the image and the sum of the pixel brightness of one of the relative straight lines of the right half of the image. The difference (absolute value) and the sum of these differences. SBS trust can reflect the similarity between the upper and lower half of the image. In general, if the image format is 3D SBS format, the SBS trust degree is small: conversely, if the image format is not the 3D SBS format, the SBS trust degree is large.

In detail, the algorithm for TB trust is as follows:

Fig. 3B is a diagram showing the TB trust degree according to the present embodiment.

Take Figure 2 as an example:

Diff _ TB ₁ = abs (SOP1-SOP9) + abs (SOP2-SOP10) + abs (SOP3-SOP11) + abs (SOP4-SOP12)

Diff _ TB ₂ = abs (SOP5-SOP13) + abs (SOP6-SOP14) + abs (SOP7-SOP15) + abs (SOP8-SOP16)

Therefore, it can be seen that, in this embodiment, the TB trust degree is the pixel brightness of one of the original blocks of one of the upper half of the image and the pixel of the original block of the lower half of the image. The difference between the sums (absolute values) and the sum of these differences. The TB trust reflects the similarity between the upper and lower half of the image. In general, if the image format is 3D TB format, its TB trust is small; conversely, if the image format is not 3D TB format, its TB trust is large.

In step 160, the format of the single frame image is detected according to the average brightness of the image, the SBS trust degree, and the TB trust degree. The details of step 160 are as shown in Figure 3B. FIG. 4 shows a detailed flowchart of detecting an image format according to the average brightness of the image, the SBS trust degree, and the TB trust degree in the embodiment.

Here, first define several parameters as follows:

In the above formula, the parameters width and height represent the width and height of the image, respectively. The parameter Mean is the average brightness of the image. In this embodiment, the introduction parameter Mean can offset the influence of the image brightness to prevent the image format judgment result from being affected by the image brightness. The parameter Mean_SBS represents the average SBS trust. The parameter Mean_TB represents the average TB trust.

In step 410, the size relationship between the parameters Mean_SBS and Mean_TB is determined. If the parameter Mean_TB is greater than Mean_SBS, it is determined whether any of the SBS signal degrees Diff_SBS _i is greater than T1*mean (step 415). T1 is the first threshold. If any of the SBS semaphores Diff_SBS _{i in the} image is greater than T1*Mean, the image is determined to be a 2D image, as shown in step 425. If it is judged to be a 2D image, it is played in the 2D playback mode.

If the parameter Mean_TB is less than Mean_SBS, it is determined whether any of the TB signal degrees Diff_TB _i is greater than T1*Mean (step 420). If any of the TB signal degrees Diff_TB _{i in the} image is greater than T1*Mean, the image is determined to be a 2D image, as shown in step 425.

If all of the SBS semaphores Diff_SBS _{i of the image} are less than T1*Mean, it is determined whether (T2*(Mean_TB) ² ) is less than (Mean_SBS) ² (step 430). If so, it means that the image format cannot be judged and belongs to an unknown format (step 440). For example, if the image is a pure subtitle (usually the subtitle only occupies a small portion of the bottom of the screen and most of the screen is black), then the difference between the SBS trust and the TB trust of the image is It may not be large, and in this embodiment, the subtitle image is regarded as an unknown format that cannot be judged. When playing such an unknown format image, the judgment result will be discarded (for example, such an unknown format image is played according to the preset play mode, or such an unknown format image is played according to the previous play mode).

Conversely, if it is judged that (T2*(Mean_TB) ² ) is not less than (Mean_SBS) ² , the SBS representing the image has a small degree of trust, and thus this image is judged to be the SBS format of the 3D format. If it is judged to be an SBS image (3D image), it is played in the SBS playback mode.

If all of the TB semaphores Diff_TB _{i of the image} are less than T1*Mean, it is determined whether (T2*(Mean_SBS) ² ) is less than (Mean_TB) ² (step 435). If so, it means that the image format cannot be judged and belongs to an unknown format (step 440). Conversely, if it is judged that (T2*(Mean_SBS) ² ) is not less than (Mean_TB) ² , the TB trustworthiness representing the image is small, and thus the image is judged to be the TB format of the 3D format. If it is judged to be a TB image (3D image), it is played in the TB playback mode.

In summary, in this embodiment, the SOP, the SBS trust degree, the TB trust degree, the image average brightness, and the like are used to automatically detect the image format, so the detection accuracy is high. In addition, the hardware implementation cost of the embodiment is also low, so it has great market potential.

In summary, although the present invention has been disclosed above by way of example, it is not intended to limit the present invention. Those who have ordinary knowledge in the technical field of the present invention can make various changes and refinements without departing from the spirit and scope of the present case. Therefore, the scope of protection of this case is subject to the definition of the scope of the patent application attached.

110~160. . . step

410~450. . . step

200. . . image

210. . . Giant block

220. . . Subblock

230. . . Original block

FIG. 1 is a flow chart showing a method for automatically detecting an image format according to an embodiment of the present invention.

Fig. 2 is a view showing a block, a sub-block and a original block according to an embodiment of the present invention.

Fig. 3A shows a schematic diagram of the SBS trust degree according to the present embodiment.

Fig. 3B is a diagram showing the TB trust degree according to the present embodiment.

FIG. 4 shows a detailed flowchart of detecting an image format according to the average brightness of the image, the SBS trust degree, and the TB trust degree in the embodiment.

110~160. . . step

Claims (22)

An automatic single-frame image format detection method includes: dividing a single-frame image into a plurality of macroblocks; dividing each of the macroblocks into a plurality of sub-blocks; and setting an original block in each of the sub-blocks; Calculating a sum of luminance values of one of the pixels of each of the original blocks; and calculating, according to the sum of the luminance values of the pixels, a first trust degree between one of the left half and the right half of the single frame image; a luminance sum total feature value, calculating a second trust degree between one of the upper half and the lower half of the single frame image; and the first trust degree of the single frame image according to the pixel luminance sum feature value and the pixel The second degree of trust detects a format of the single frame image. The method of claim 1, wherein the step of dividing the single-frame image into a plurality of macroblocks comprises: dividing the single-frame image horizontally and vertically. The method of claim 1, wherein the step of dividing each of the plurality of chunks into a plurality of sub-blocks comprises: dividing each of the chunks horizontally and vertically. The method of claim 1, wherein an upper left corner of one of the original blocks is aligned with an upper left corner of the sub-block. The method of claim 1, wherein: one of the original blocks has a width between 1 and a width of the sub-block; and a height of the original block is between 1 and a height of the sub-block between. The method of claim 1, wherein the pixel luminance sum characteristic value is expressed as: Where Luma represents a pixel brightness, i represents a label of the i-th macro block, j represents the number of the j-th sub-block within the i-th macro block, N _Marco represents the total number of the macro blocks, and N _Sub represents The total number of the sub-blocks in each chunk, SOP _{i , j} represents the sum of the luminances of all the pixels of the j-th sub-block of the i-th chunk. The application method of claim 6 patentable scope of the item, wherein the calculating the first trust Diff image of the SBS _m _ step comprises: Among them, abs represents the absolute value. The application method of claim 7 patentable scope item, wherein calculating the second image of the trust Diff TB _m _ steps of comprising: . The method of claim 8, wherein the step of detecting the format of the single frame image comprises: comparing a first average of the first trust degree of the single frame image with the second trust a second average value; if the first average value is less than the second average value, and the first trust level is greater than a product of a first threshold value and an image average brightness characteristic value, determining the single frame image Is a two-dimensional image; if the first average value is greater than the second average value, and the second trust level is greater than the product of the first threshold value and the image average brightness characteristic value, determining that the single frame image is one a two-dimensional image; if the first average value is smaller than the second average value, all first trust degrees of the single-frame image are smaller than the product of the first threshold value and the image average brightness characteristic value, and T2*(Mean_TB ^{2 is} less than (Mean_SBS) ² , it is determined that the format of the single frame image is unknown, wherein T2 represents a second threshold, Mean_SBS represents the first average, and Mean_TB represents the second average; if the first average The value is less than the second average All the first image of a single frame of confidence is less than the first threshold value with the product of the average luminance characteristic of the image, and T2 * (Mean_TB) ² is greater than (Mean_SBS) ^2, it is determined that the format of the image of a single frame a left and right side-by-side format of the three-dimensional format; if the first average value is greater than the second average value, all second trust degrees of the single-frame image are smaller than the product of the first threshold value and the image average brightness characteristic value, and T2 *(Mean_SBS) ^{2 is} less than (Mean_TB) ² , it is determined that the format of the single frame image is unknown; and if the first average value is greater than the second average value, all second trust degrees of the single frame image are smaller than the first A product of the threshold value and the average brightness characteristic value of the image, and T2*(Mean_SBS) ^{2 is} greater than (Mean_TB) ² , determining that the format of the single-frame image is a top-down side-by-side format of the three-dimensional format. A single-frame video playback method includes: dividing a single-frame image into a plurality of macroblocks; dividing each of the macroblocks into a plurality of sub-blocks; and setting an original block in each of the sub-blocks; And determining, according to the sum of the pixel brightness, a first half of the single frame image is similar to a second half, and if yes, determining the single frame The image is in a three-dimensional format; and the single-frame image is played according to the determined image format. The method of claim 10, wherein the step of dividing the single-frame image into a plurality of macroblocks comprises: dividing the image horizontally and vertically. The method of claim 10, wherein the step of dividing each of the plurality of chunks into a plurality of sub-blocks comprises: dividing each of the chunks horizontally and vertically. The method of claim 10, wherein an upper left corner of one of the original blocks is aligned with an upper left corner of the sub-block. The method of claim 10, wherein: one of the original blocks has a width between 1 and a width of the sub-block; and a height of the original block is between 1 and a height of the sub-block between. The method of claim 10, wherein determining whether the left half of the single-frame image is similar to a right half according to the sum of the luminance values of the pixels, and if so, determining that the single-frame image is A side-by-side format of a three-dimensional format. The method of claim 15, wherein determining whether the upper half of the single frame image is similar to the lower half according to the sum of the luminance characteristic values of the pixels, if yes, determining that the single frame image is one The top and bottom side format of the 3D format. The method of claim 16, wherein the pixel luminance sum characteristic value is expressed as: Where Luma represents a pixel brightness, i represents a label of the i-th macro block, j represents the number of the j-th sub-block within the i-th macro block, N _Marco represents the total number of the macro blocks, and N _Sub represents The total number of the sub-blocks in each chunk, SOP _{i , j} represents the sum of the luminances of all the pixels of the j-th sub-block of the i-th chunk. The method of claim 17, further comprising: calculating a first trust degree of the image Diff _ SBS _m , Diff _ SBS _m = Abs represents an absolute value; and calculates a second trust degree Diff _ TB _{m of} the image, The method of claim 18, wherein the step of determining that the single frame image is the left and right side-by-side format comprises: if the first average value of the first trust degree of the single frame image is smaller than the second a second average value of the trust, all first trust degrees of the single frame image are less than a product of a first threshold value and an image average brightness characteristic value, and T2*(Mean_TB) ^{2 is} greater than (Mean_SBS) ² , Then, the format of the single-frame image is determined to be a left-right side-by-side format in a three-dimensional format, where T2 represents a second threshold value, Mean_SBS represents the first average value, and Mean_TB represents the second average value. The method of claim 19, wherein the step of determining that the single frame image is in the up and down side by side format comprises: if the first average value is greater than the second average value, all second of the single frame image The trust degree is less than the product of the first threshold value and the average brightness characteristic value of the image, and T2*(Mean_SBS) ^{2 is} greater than (Mean_TB) ² , and the format of the single frame image is determined to be the upper and lower side-by-side format of the three-dimensional format. The method of claim 20, further comprising: if the first average value is less than the second average value, and the first trust level is greater than the product of the first threshold value and the image average brightness characteristic value And determining, the single frame image is a two-dimensional image; and if the first average value is greater than the second average value, and the second trust level is greater than the product of the first threshold value and the image average brightness characteristic value, Then, the single frame image is determined to be a two-dimensional image. The method of claim 21, further comprising: if the first average value is less than the second average value, all first trust degrees of the single frame image are less than the first threshold value and the average brightness of the image a product of the feature values, and T2*(Mean_TB) ^{2 is} less than (Mean_SBS) ² , determining that the format of the single frame image is unknown; and if the first average value is greater than the second average value, all of the single frame image The second trust degree is less than the product of the first threshold value and the image average brightness characteristic value, and T2*(Mean_SBS) ^{2 is} less than (Mean_TB) ² , and the format of the single frame image is determined to be unknown.