KR20150057026A - Method for Detecting a Logo in Moving Image In Real Time basis - Google Patents

Abstract

The present invention relates to a method for detecting a logo in a video on a real-time basis. According to the present invention, the video logo detection method includes: a first step of detecting edge pixels present in the respective video frames continuously used to display a video and allocating the edge data to the pixels of the video frames to distinguish the pixels from other pixels; a second step of sequentially accumulating the edge data of the respective video frames in the same pixel locations to calculate the sum data which is the accumulated edge data for the respective pixels, and generating an edge map including the sum data from all of the pixels in the video frames; and a third step of determining the pixels with the sum data exceeding a threshold value as a logo area when the sum data in the edge map exceeds the threshold value.

Description

[0001] The present invention relates to a method for detecting a real-

The present invention relates to a real-time logo detection method for moving images.

A television signal displays a logo on a frame of a moving image in order to identify the broadcasting station or guide the broadcasting program. However, for viewers, these logos are disturbing and inconvenient for viewing video. Particularly, since such a logo is generally expressed at a high luminance, it also causes a problem of deterioration of certain pixels in a region where the logo is located.

In order to solve these problems, methods of detecting and removing a logo from a video image have been proposed. For example, U.S. Patent No. 5,668,917 detects logos based on the repetitive nature of opaque logos. And, an opaque logo detection apparatus and method in a digital video signal disclosed in Korean Patent Publication No. 10-2005-0034538 discloses a technique of detecting a logo by comparing key frames having a scene change.

These prior arts are based on comparing pixel data between image frames as a basic method for detecting a logo. That is, in order to implement the conventional logo detection method, at least two frame memories are required to store image frames to be compared. That is, the cost required to implement the method increases, and the configuration becomes complex to use the frame memory.

Also, since the prior art detects the logo through a comparison between image frames, there is a disadvantage that it takes a lot of time for logo detection at every moment.

It is an object of the present invention to provide a real-time logotype detection method of a moving image which can reduce a logo detection time while simplifying the configuration.

A method of detecting a logo of a moving image according to the present invention is a method for detecting an edge pixel existing in each image frame in a plurality of consecutive image frames for representing a moving image and separating the pixels of the image frames from other pixels The edge data of each of the image frames is sequentially accumulated for each of the same pixel positions to calculate summation data which is accumulated edge data for each of the pixels, A second step of generating an edge map including the summation data for the edge area, and a third step of, when the summation data exceeds the threshold in the edge map, .

The method of detecting a moving image according to the present invention detects a logo using edges of image frames received in real time and corrects the image with respect to a logo area of the detected image. That is, since the present invention can perform projection correction for each single image frame after the time corresponding to the reference time in the first frame, real-time image correction is possible.

In addition, according to the method of detecting a logo of a moving picture according to the present invention, since one frame memory and a line memory are used instead of requiring two or more frame memories for storing image frames, Can be simplified.

1 is a block diagram showing a configuration of a moving image logo detection device according to the present invention;
2 is a flowchart showing a method of detecting a logo of moving images according to the present invention;
3 to 9 are views showing an embodiment of a method of detecting a logo of moving images according to the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Reference will now be made in detail to the preferred embodiments of the present invention, examples of which are illustrated in the accompanying drawings. Like reference numerals throughout the specification denote substantially identical components. In the following description, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

A logo detection apparatus according to the present invention is for detecting a logo on a screen of a moving picture displayed on a display device. A moving picture represents motion based on continuously displaying a plurality of video frames on a display device, and the moving picture may be reproduced on the basis of a broadcasting signal transmitted from a broadcasting station, or a video file stored in a storage medium have. The logo is continuously displayed for a predetermined time or longer at a predetermined position of the image, such as the title of the corresponding image displayed in the moving image or the producer or distributor of the corresponding image.

The logo detection apparatus according to the present invention detects a pixel for displaying a logo in a moving image and corrects the brightness of the corresponding pixel. The logo detection apparatus may be included in a T-con of the display apparatus, It is possible. The present invention can be applied to an OLED (Organic Light Emitting Display) for preventing deterioration of image quality due to deterioration of a specific pixel in a panel. However, in order to improve the inconvenience of viewing a video image of a viewer due to a logo And can also be applied to a display device such as a liquid crystal display (LCD), a field emission display (FED), a plasma display panel (PDP), and an electrophoresis (EPD) It is self-evident.

1 is a block diagram showing a configuration of a logo detection apparatus 100 according to the present invention.

1, a logo detection apparatus 100 according to the present invention includes an edge detection unit 110, an edge data summation unit 120, a noise detection unit 130, a logo detection unit 140, and an image correction unit 150 Respectively.

The edge detecting unit 110 continuously receives a plurality of image frames for representing a moving image and detects edge pixels included in each image frame. The edge detecting unit 110 detects whether each of the pixels is an edge pixel Is added to the edge data.

The edge detecting unit 110 may determine whether an arbitrary pixel is an edge pixel based on a luminance difference between pixels in a method for detecting an edge pixel. For this purpose, the edge detector 110 calculates the luminance difference between adjacent pixels for all the pixels using the edge mask 112. The edge detector 110 determines whether the luminance difference between pixels is equal to or greater than the luminance variation threshold. If the luminance of an arbitrary pixel is greater than the luminance of the adjacent pixel and the luminance is equal to or higher than the luminance reference value, the edge detecting unit 110 gives edge data of '1' to the pixel. Or if there is a luminance difference between the luminance of an arbitrary pixel and the luminance of the adjacent pixel and the luminance variation threshold, the edge detector 110 gives edge data of '0' to the pixel.

In this process, the luminance change threshold for discriminating the edge data may be set to a value that can be used as a criterion for distinguishing whether the luminance difference between adjacent pixels is different from each other. For example, the luminance variation threshold may be set to '20' when the luminance of an image is represented by 256 luminance.

The mask used by the edge detection unit 110 may be a 3 × 3 mask. The edge detection unit 110 may use a 5 × 5 mask or a 10 × 10 mask, but a 3 × 3 mask can be used to reduce the line memory.

As a practical implementation, the edge detector 110 of the present invention may utilize line memory. Therefore, the edge detection method according to the present invention can reduce the cost for implementing the invention by reducing the frame memory.

The edge data summation unit 120 accumulates and stores the edge data obtained by the edge detection unit 110 in the frame memory 10, thereby generating an edge map including the sum data, which is accumulated edge data. At this time, the edge data summing unit 120 counts the number of times the edge data of '1' is accumulated.

In order to generate the edge map, the edge data summation unit 120 generates the first edge map based on the first edge data of the first frame obtained by the edge detection unit 110. The edge data summing unit 120 sums the second edge data of the pixels obtained in the second frame with the first edge data of the pixels of the first edge map, 2 Create an edge map.

At this time, in the process of summing edge data of each pixel of the edge data summing unit 120, only the edge data of '1' is added to the sum data of the previous frame, and for pixels having edge data of '0' And initializes the sum data. For example, when the sum data up to the (i-1) -th frame is assumed to be 'k', the edge data summation unit 120 adds the edge data of the i- K + 1 ". If the edge data of the i-th frame is '0', the edge data adder 120 initializes the sum data of the ith edge map to '0'.

In this way, the edge data summation unit 120 can generate the n-th edge map for the n-th frame in the first frame.

The edge data summing unit 120 counts the number of times the edge data of '1' is added, and checks whether the sum data exceeds the threshold value. The process of checking whether the edge data summing unit 120 has exceeded the threshold value of the summation data is for judging whether it is an estimated logo pixel. That is, the larger the sum data, the more the edge of a certain size or more is continuously displayed in the corresponding pixel. This is because the pixel is the estimated logo pixel which is likely to be the logo region. The threshold value is for comparing the summation data and is associated with a consecutive number of frames. That is, the threshold value is closely related to the reference time for estimating the logo in the image. Therefore, it is set considering the duration of the logo in the image. If the reference time is small, there is a high possibility that the estimated logo pixel is detected by the image pattern in addition to the logo, and if the reference time is long, it may take a long time to detect the first logo. Therefore, the reference time can be set within a range in which the logo detection probability can be increased within a range where the time required for logo detection is not long. For example, the reference time may be set to about two minutes. Therefore, in an image having 60 frames per second, the threshold value can be set based on a criterion of 60 × 120 (sec) = 7200 (frame). That is, the edge data summation unit 120 can detect the estimated logo pixel by detecting that the sum data exceeds the set threshold value in the range of about 7200, for example, in the range of 6000 to 8400.

The noise detection unit 130 detects and removes noise from the edge map generated by the edge data summation unit 120.

The operation in which the noise detector 130 detects noise is performed based on the fact that the edge detector 110 has confirmed the estimated logo pixel. That is, when the edge detector 110 detects the estimated logo pixel, the noise detector 130 determines whether the estimated logo pixel is a noise or a logo area.

The noise detector 130 can determine whether or not the estimated logo pixels are noise based on the continuously sized size. Since the logo in the image is intended to be displayed visually to the viewer, it is clearly distinguished from the impulsive noise (Impluse Noise) which is several pixels. Accordingly, the noise detector 130 regards the estimated logo pixel, which is composed of several pixels or dozens of pixels, as noise and removes it. For this purpose, a median filter can be used.

The logo detection unit 140 detects substantial logo pixels based on the estimated logo pixels from which the noise is removed through the noise detection unit 130. The estimated logo pixels are detected based on the sum of the edge data. Consequently, the estimated logo pixel is the result of detecting only the edge, that is, the boundary line of the object. Accordingly, the logo detection unit 140 detects a substantial logo area including the inside of the border lines. For this, the logo detection unit 140 may use a scan-line algorithm.

The image correcting unit 150 performs image correction on the data of the logo pixels included in the logo area detected by the logo detecting unit 140. The logo data detected by the logo detecting unit 140 is pixel data having a high luminance continuously. To compensate for this, the image correcting unit 150 performs a process of lowering the brightness of the logo data. At this time, if the brightness of the logo pixels is excessively lowered, the display quality may be deteriorated because the image is displayed in the dark, and if the brightness is lowered too much, it may be insufficient to improve the device deterioration phenomenon of the panel. Accordingly, the logo detecting unit 140 performs image correction that lowers the brightness of 40 to 60% in the data received in the logo area.

A method of detecting a logo according to the present invention using the above-described logotype detecting apparatus will be described with reference to the flowchart of FIG.

≪ Edge data acquisition through edge detection: S201 >

In order to detect the logo, the edge detection unit 110 sequentially receives image frames for displaying moving images. The edge detection unit 110 detects edge pixels in an image frame sequentially received and adds edge data.

3 is a diagram showing a part of an image frame in which a character logo 'OCN' is displayed. A method of detecting the logo L when the bright logo L is expressed in a dark scenery will be described with reference to the 'A' region as follows

4 is an enlarged view of the area 'A' shown in FIG. 3 and a schematic diagram showing the brightness of pixels belonging to the [j + 1] th row.

In order to detect an edge pixel, the edge detecting unit 110 uses the edge mask 112 to determine the center pixel as an edge when the luminance difference between adjacent rows or adjacent columns is equal to or larger than a predetermined magnitude. For example, the edge mask 112 sequentially detects the edge pixels of each row while moving from the left column to the right column sequentially. When edge pixels are detected for one row, edge detection for the next row is performed to detect edges of all pixels of the image frame.

To confirm that each pixel is an edge pixel, the edge mask 112 calculates the luminance difference between the right adjacent pixel and the center pixel C _ij of the edge mask 112. The edge mask 112 can detect the center pixel of the edge mask 112 as an edge pixel when the calculated luminance difference is equal to or greater than the luminance reference value. For example, when the center pixel C _ij of the edge mask 112 is located at the 'P9' pixel, the luminance difference d 'between the pixel P9' and the pixel at the right adjacent pixel (j + 1, K + ) Exceeds the threshold value, the edge mask 112 can detect the pixel of 'P9' as an edge pixel. Likewise, the luminance difference d between the pixel P10 'and the pixel' 16 'with respect to the right adjacent pixel is determined to be equal to or greater than the luminance reference value, and is detected as an edge.

Similarly, when the center pixel C _ij of the edge mask 112 is located at the pixel 'P 1', the pixel of (j + 1, K + 1) As shown in Fig. Accordingly, the edge mask 112 can detect the pixel 'P1' as an edge pixel. Likewise, the 'P2' to 'P8' pixels are also detected as edges by discriminating the luminance difference between the neighboring pixels on the lower side and the luminance reference value.

The edge detection unit 110 adds edge data to the pixels P1 to P16 detected by the edges as '1'. In addition, the edge detection unit 110 adds '0' edge data to pixels not detected as edges. At this time, since edge data of '0' and '1' are used to discriminate whether or not edge pixels are included, edge data of different sizes may be added.

<Edge map generation through accumulation of edge data: S203>

The edge data summation unit 120 accumulates and stores the edge data acquired by the edge detection unit 110 in the frame memory 10, thereby generating an edge map.

For example, when the first frame F1 shown in FIG. 4 is the first image frame, the edge data adder 120 adds edge data of '1' added to the pixels P1 to P16 detected as edges And stores it in the frame memory 10.

5 is a diagram showing an edge map of the 'A' region of the first frame F1 shown in FIG. That is, in the first edge map M1, only the pixels 'P1' to 'P16' in the 'A' region have data of '1', and the remaining pixels have the data of '0'.

In this way, the edge data summation unit 120 generates the first edge map by storing the edge data of '1' for the pixels detected as edges in the first frame P1 in the frame memory 10.

The edge data of the first edge map is regarded as the first sum data.

Next, the edge data summation unit 120 generates the second edge map including the summation data by adding the edge data of the second frame to the first edge data to generate summation data. That is, if the 'A' area of the second frame following the first frame F1 is the same as the 'A' area of the first frame F1 shown in FIG. 4, The edge data of '1' added to the pixels P 1 to P 16 is added to the edge data of the first edge map. Accordingly, the second edge map M2 of the 'A' region up to the second frame appears as shown in FIG.

In this way, the edge data summation unit 120 generates the second summation data by summing the edge data up to the second frame, thereby generating the second edge map M2 including the second summation data.

Then, the edge data summation unit 120 initializes the summation data up to the previous value to '0' for pixels having edge data of '0' in a certain frame. For example, as shown in FIG. 7, when the logo 30 disappears in the third frame F3 and the pixels of P1 'to P16' are not detected as edges, the edge data summation unit 120 adds 'P1' For the pixels of 'P16', the sum data of '2' included in the second edge map is initialized to '0'. Therefore, the 'A' area of the third edge map M3 including the sum data up to the third frame F3 appears as shown in FIG.

<Estimated logo pixel detection by counting the sum data: S205 and S207>

The edge data summing unit 120 accumulates the edge data and compares the sum data of pixels stored in the frame memory 10 with a threshold value in step S203. As in the embodiment, when the edge data for the edge pixels is given as '1', the edge data summation unit 120 can calculate the size of the summation data by counting edge data of '1' have.

When the sum data exceeds the threshold value as a result of the comparison between the sum data and the threshold value, the corresponding pixel is regarded as the estimated logo pixel.

If the sum data is less than or equal to the threshold value, the edge data summing unit 120 repeats steps S201 and S203.

<Noise detection and removal: S209>

When it is determined in step S205 that the sum data counted by the edge data summation unit 120 exceeds the threshold value, the noise detection unit 130 checks whether the estimated logo pixel is noise, and removes the identified noise. For this, the noise detector 130 may filter several or tens of consecutive pixels as impulsive noise by filtering the edge map including the sum data using the median filter.

<Logo detection: S211>

In step S207, the estimated logo pixels from which the noise has been removed are regarded as the edges of the logo. Then, the logo detection unit 140 detects a substantial logo area within the border area. For this purpose, the logo detection unit 140 may use a scan-line polygon fill algorithm (Scan-Line Polygon Fill Algorithm).

FIG. 9 is a diagram illustrating an embodiment of the logo detection unit 140, which shows that logo detection up to the 'k' scan line is performed. In this manner, when the scan line passes the pixels, the logo detecting unit 140 detects a point intersecting the edge pixels of the logo and repeats the process of filling the space between the edge pixels. In this way, the logo detection unit 140 detects the logo 30 by filling pixels corresponding to the inner area between the edge pixels.

&Lt; Image correction: S213 >

If a substantial logo region is detected in step S209, the image correction unit 150 performs a correction to lower the brightness of the pixels included in the logo region in the image frame. That is, the process of removing the noise in step S207 and detecting the logo area in step S209 is performed using the edge map, but the image correction unit 150 performs the image correction on the image frame received by the edge detection unit 110 do.

For example, if the i-th accumulated data of the edge data of the i-th frame provided by the edge detecting unit 110 exceeds the threshold value and steps S207 and S209 proceed, the image correcting unit 150 calculates And corrects the image in the logo region pixel.

At this time, the image correcting unit 150 can perform image correction to lower the brightness in the logo region pixel.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the invention. Therefore, the technical scope of the present invention should not be limited to the contents described in the detailed description of the specification, but should be defined by the claims.

30: Logo 112: edge mask

Claims (9)

A first step of detecting an edge pixel existing in each of the plurality of continuous image frames for representing a moving image and giving edge data distinguished from other pixels to pixels of the image frames;
Wherein the edge data of each of the image frames is sequentially accumulated for each of the same pixel positions to calculate summation data which is accumulated edge data for each of the pixels and an edge including the summation data for all pixels of the image frames A second step of generating a map; And
And determining that the pixel having the summation data exceeding the threshold is a logo region when the sum data exceeds the threshold value in the edge map.
The method according to claim 1,
In the first step,
Wherein when the luminance of the center pixel of the edge mask is equal to or greater than the luminance of the adjacent pixel and the luminance reference value, the edge pixel center pixel is determined as an edge pixel Wherein the detecting step detects the presence of the real-time logo.
The method according to claim 1,
The second step is to generate edge maps for the first to n &lt; th &gt; image frames continuously provided,
Generating a first edge map by considering edge data of each pixel of the first image frame as first sum data;
Summing edge data of the second image frame to the first edge map to generate a second edge map including second sum data; And
And generating an n-th edge map by summing edge data of the n-th image frame in an (n-1) -th edge map, wherein n is a natural number of 3 or more ).
The method of claim 3,
The second step
Edge data is added to the sum data of the corresponding pixels of the (i-1) -th edge map when the edge data of any pixel of the i-th edge map is '1'
And when the edge data of a certain pixel of the i-th edge map is '0', the sum data of the corresponding pixel of the (i-1) -th edge map is initialized to '0' Detection method (i is a natural number of 2 or more).
The method of claim 3,
In the third step,
And when the i-th sum data of an arbitrary pixel in the i-th edge map exceeds the threshold value set in a range of 6000 to 8400, the pixel is judged to be a logo region. , i is a natural number of 2 or more).
6. The method according to any one of claims 1 to 5,
After the third step,
And a fourth step of detecting a noise pixel among the pixels determined to be the logo region.
7. The method of claim 6,
After the fourth step,
Further comprising a fifth step of initializing the sum data of the noise pixels to '0' for the pixel detected as the noise pixel.
7. The method of claim 6,
After the fourth step,
And a fifth step of correcting the image of the pixels determined to be the logo area.
9. The method of claim 8,
The fifth step
And the brightness of pixels of the logo area detected in the edge map is reduced by 40 to 60%.

Images