JP2000187732A - Method and device for deciding character area and recording medium storing the method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the discrimination accuracy of each character area in an image including characters. SOLUTION: A color image input storage part 1 inputs color image data including characters and stores the inputted data in a memory. An image area division part 2 divides the inputted and stored color image into connected pixel areas by using a previously determined method. A high contrast area deciding part 3 discriminates a character area highly accurately by executing processing for detecting an edge from an area boundary part between plural divided areas, leaving an area in which the ratio of peripheral length to edge pixels is more than a threshold as a high contrast area having high possibility of including a character and disusing other areas as low contrast areas and then stores the obtained character area image in an character area image storing part 4.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to, for example, a plurality of frame images constituting a color moving image such as a video image of a television broadcast, among a frame image in which characters such as telop characters in the television broadcast image are displayed. The present invention relates to a character region determination technology for extracting a character portion from a character region as a pixel connection region.

[0002]

2. Description of the Related Art A character region determination for extracting a character portion as a pixel connection region from a frame image in which characters are displayed in a plurality of frame images constituting a color moving image or a still color image in which characters are displayed. A great deal of research has been conducted on technology.

Reference [1]: Kuwano, Kurakake, Odaka: "A telop character extraction method for video data search", IEICE Technical Report, PRMU 96-98, pp. 139-143. 39-46, (199
The method proposed in 6-11) (hereinafter referred to as conventional method [1]) divides an input color image into a connected pixel area by a color space division process,
It is determined whether or not the obtained area is a character area. In the conventional method [1], as shown on the left side of FIG. 6, a region whose area value is equal to or less than a certain value, a region that does not touch the image frame, and a region whose position does not change for a certain time are determined as character regions.

[0004]

However, in the conventional method [1], as shown in the right side of FIG. 6, in the character area determination processing after the area division processing, the area has about the same area as the character, and There is a problem that a region other than the character which does not touch the frame and whose position does not change for a certain period of time remains.

[0005] Therefore, the present invention provides a method for removing characters other than characters, for example, in a video, which has the same area as the characters, does not touch the frame of the image, and does not change its position for a certain period of time. It is an object of the present invention to provide a method and an apparatus for determining a character region in which the accuracy of the determination of a character region in an image including the character region is improved.

[0006]

In order to solve the above-mentioned problems, a character area extracting method according to the present invention comprises a first step of inputting an image in which characters are displayed and storing the image as an original image; A second step of dividing the original image input and stored in the first step into connected pixel regions by using a predetermined method to obtain a region-divided image, and a region dividing process obtained in the second stage. For each region in the image, a contrast feature representing the difference in brightness between the inside of the region and the outside of the region at the region boundary is calculated, and a region where the contrast feature is larger than a preset value is set as a character region and the region divided image is calculated. A third step of deleting the remaining area from the area divided image from the area divided image and storing a character area image including all the connected pixel areas determined as the character area obtained in the third step. 4th
And a step of:

Alternatively, in the above-described character region extraction method, the third step is to use an edge pixel in the image by using a predetermined method (for example, using a Robinson edge detection operator) on the input original image. 3-1), the number of boundary pixels inside each region in the region-divided image is calculated, and the number of boundary pixels inside each region in the region-divided image is calculated. Among them, a third stage of calculating the number of pixels which are edges detected in the third stage, a number of boundary pixels inside each region obtained in the third stage, A third-fourth step of calculating the ratio of the number of pixels that are edges among boundary pixels inside each region obtained in the third-third step; The number of boundary pixels inside the If the ratio of the number of pixels that are the same is greater than a preset value, the region is determined as a character region and is left in the region-divided image; otherwise, the region is determined as a background noise region. And a 3-5th step of deleting from the region divided image.

Further, a program for causing a computer to execute the steps in the above character area determination method is recorded on a recording medium readable by the computer.

According to another aspect of the present invention, there is provided a character region extracting apparatus comprising: an image input storage unit for inputting an image in which characters are displayed and storing the input image as an original image; The input and stored original image is divided into connected pixel areas using a predetermined method,
Image region dividing means for obtaining a region divided image, and for each region in the region divided image obtained by the image region dividing means,
A contrast feature representing the difference in brightness between the inside of the region and the outside of the region at the region boundary is calculated, and a region in which the contrast feature is larger than a predetermined value is left as a character region in the region-divided image. A high-contrast region determining unit that deletes from the region-divided image; and a character-region image storing unit that stores a character region image including all connected pixel regions determined as character regions obtained by the high-contrast region determining unit. , The image input storage means,
A control unit for controlling an execution order of the image area dividing means, the high contrast area determining means and the character area image storing means.

Alternatively, in the above-described character region extracting apparatus, the high-contrast region determining means detects an edge pixel in the image by using a predetermined method (for example, using a Robinson edge detection operator) on the original image. Edge detecting means, and a region perimeter calculating means for calculating the number of boundary pixels inside each region in the region divided image,
Area boundary edge calculation means for calculating the number of pixels which are edges detected by the edge detection means among boundary pixels inside each area in the area divided image; Area perimeter / edge ratio calculating means for calculating a ratio between the number of boundary pixels inside the area and the number of pixels which are edges among the boundary pixels inside each area obtained by the area boundary edge calculating means; If the ratio of the number of boundary pixels inside the region obtained by the region perimeter / edge ratio calculation means to the number of edge pixels among the boundary pixels inside the region is larger than a preset value, Determine the area as a character area and leave it in the area divided image,
If not, a character region determining unit that determines the region as a background noise region and deletes the region from the region divided image is provided.

In general, a character displayed in an image has a feature that the luminance value contrast with the surroundings is often high. In the conventional method [1], since the brightness contrast around the area is not evaluated, the area has about the same area as the character,
A region of a general object having a low brightness contrast at a boundary portion among regions other than characters that do not touch the frame of the image and whose position does not change for a certain period of time remains.

Therefore, according to the present invention, after the region is divided, the contrast characteristic of the brightness value at the boundary of each region is calculated,
By leaving only a high contrast area, it is possible to improve the accuracy of character area determination.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration and processing flow of an apparatus according to an embodiment of the present invention.

In FIG. 1, reference numeral 1 denotes a color image input storage unit, which inputs image data such as a color image in which characters are displayed and stores it in a memory.

Reference numeral 2 denotes an image area dividing unit which determines an original image such as a color image input and stored in the color image input / storage unit 1 (for example, a method proposed by the conventional method [1], etc.). ) To divide into connected pixel regions.

Reference numeral 3 denotes a high-contrast region determining unit which calculates a contrast characteristic of a luminance value at a region boundary for each region in the region-divided image obtained by the image region dividing unit 2, and sets the contrast characteristic in advance. An area larger than the set value is determined as a character area.

Reference numeral 4 denotes a character area image storage unit, which stores the character area image obtained by the high contrast area determination unit.

Reference numeral 5 denotes a processing control unit which controls the execution order of each of the above-mentioned units 1-4.

FIG. 2 is a block diagram showing an example of the processing flow together with an example of the configuration of the high-contrast area determination section 3 in FIG.

In FIG. 2, reference numeral 6 denotes an edge detector.
A predetermined method (for example, Rob
using an inson edge detection operator, etc.)
Is used to detect edge pixels in the image.

Reference numeral 7 denotes an area perimeter calculating unit which calculates the number of boundary pixels inside each area in the area divided image.

Reference numeral 8 denotes a region boundary edge calculation unit which calculates the number of edge pixels obtained by the edge detection unit 6 among boundary pixels inside each region in the region division image.

Reference numeral 9 denotes an area perimeter / edge ratio calculation unit.
The ratio between the number of boundary pixels inside each region obtained by the region perimeter calculation unit 7 and the number of edge pixels among the boundary pixels inside each region obtained by the region boundary edge calculation unit 8 is calculated.

Numeral 10 denotes a character area judging section. If the ratio of the total perimeter to the number of edge pixels on the inner boundary of the area obtained by the area perimeter / edge ratio calculating section 9 is larger than a predetermined value, The region is determined as a character region and is left in the region-divided image. Otherwise, the region is deleted from the region-divided image as a background noise region.

Reference numeral 11 denotes a processing control unit for controlling the execution order of the units 6 to 10. The processing control unit 11 may be configured to also serve as the processing control unit 5 in FIG.

FIG. 3 is a flow chart for explaining an embodiment of the processing of the high contrast area determination section 3 in FIG.

In FIG. 3, step (31) is a process of reading the color original image and the region-divided image into the memory.

Step (32) is a process of detecting an edge pixel from the original image by using a predetermined method (for example, by using a Robinson edge detection operator).

Step (33) is a process of initializing a variable n for the number of regions in the region-divided image to one.

In step (34), the variable n is compared with the total number N of the areas in the divided image. If the value of n is equal to or less than N, the process proceeds to step (35). This is the process of ending the processing.

In step (35), n in the divided image
For the second region, the number P of pixels on the inner boundary of the region
This is the process of calculating (n).

In step (36), the number Q (n) of edge pixels obtained in step (32) among the inner boundary pixels of the n-th area obtained in step (35)
Is the process of calculating

Step (37) consists of P (n) and Q (n) obtained in steps (35) and (36).
Is calculated, and if the obtained value is equal to or larger than a preset value, the process proceeds to step (38); otherwise, the process proceeds to step (39).

Step (38) consists of P (n) and Q (n)
Is greater than or equal to a preset value, n
This is a process in which the second region is determined as a character region and is left in the region divided image.

Step (39) consists of P (n) and Q (n)
If the ratio is smaller than a preset value, the n-th region in the region-divided image is determined as a background noise region, and is deleted from the region-divided image.

Step (40) is a process of returning to step (34) after incrementing the variable n by one.

FIG. 4 is a diagram showing the effect of the present invention. The main feature of the present invention is that, in a process of extracting a character region from an original image such as a video including characters, the original image is divided into regions, and edges are detected from region boundaries (FIG. 4).
(A)) performing a process of leaving an area in which the ratio between the perimeter and the edge pixel is equal to or greater than a threshold value as a high-contrast area having a high possibility of including a character, and discarding other areas as a low-contrast area. Accordingly, it is possible to delete a region other than a character (for example, a low-contrast still image region) and leave only a character region with high accuracy (FIG. 4).
(B)). This makes it possible to improve the accuracy of character recognition. Such a characteristic configuration is the processing in each of the units 6 to 10 in FIG. 2, and the details of the processing are the processing in steps (35) to (39) in FIG.

In the above embodiment, the case where the present invention is applied to a color image as an original image has been described as an example.
The present invention can be applied to both cases of (1) an image whose input is binarized and (2) an ordinary image which is not binarized.

FIG. 5 shows that the present invention is most effective.
As shown in Fig. 7, the "character area extraction processing" is performed as the preprocessing, the "character area determination processing" of the present invention is performed, and then the character recognition processing is further performed.

It should be noted that some or all of the means shown in FIGS. 1 and 2 can be made to function using a computer, or that the steps (steps) of the processing shown in FIGS. 1, 2 and 3 can be performed. Needless to say, the program can be executed by a computer, a program for causing a computer to function as the means, or a program for causing a computer to execute the steps of the processing, a computer-readable recording medium, for example, FD (floppy disk), MO, ROM,
It can be recorded on a memory card, a CD, a DVD, a removable disk, etc., provided, and distributed.

[0042]

As described above, according to the present invention,
According to the present invention, the contrast characteristic of the luminance value at the boundary between the regions is calculated after the region is divided, and only the region with high contrast is left, so that the effect of improving the accuracy of determining the character region can be obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration and a processing flow of a character area determination device according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating an example of a processing flow together with an example of a configuration of a high-contrast area determination unit in the embodiment.

FIG. 3 is a flowchart illustrating an example of processing performed by a high-contrast area determination unit 3 in FIG. 1;

FIGS. 4A and 4B are diagrams illustrating an example of a character area determination result according to the embodiment of the present invention.

FIG. 5 is a diagram illustrating an application example in which the effects of the present invention are best exhibited.

FIG. 6 is a diagram illustrating an example of a character area determination result according to a conventional method [1].

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Color image input storage part 2 ... Image area division part 3 ... High contrast area determination part 4 ... Character area image storage part 5 ... Processing control part 6 ... Edge detection part 7 ... Area perimeter calculation part 8 ... Area boundary edge calculation Unit 9: Area perimeter / edge ratio calculation unit 10: Character area determination unit 11: Processing control unit

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Shoji Kurakake 3-19-2 Nishi-Shinjuku, Shinjuku-ku, Tokyo Japan Telegraph and Telephone Corporation (72) Inventor Toshiaki Sugimura 3- 192-1 Nishi-Shinjuku, Shinjuku-ku, Tokyo No. Nippon Telegraph and Telephone Co., Ltd. F-term (reference)

Claims (5)

[Claims] 1. A first step of inputting an image displaying characters and storing it as an original image, and connecting the original image input and stored in the first step using a predetermined method. A second step of dividing the image into pixel areas to obtain an area-divided image; and, for each area in the area-divided image obtained in the second step, the contrast between the inside of the area at the area boundary and the outside of the area. A third step of calculating a contrast feature representing a difference between the two, leaving an area in which the contrast feature is larger than a preset value as a character area in the area divided image, and deleting the other area from the area divided image, A fourth step of storing a character area image including all the connected pixel areas determined as the character area obtained in the third step. 2. The character area extracting method according to claim 1, wherein the third step is a step of detecting an edge pixel in the input original image by using a predetermined method. A third step of calculating the number of boundary pixels inside each region in the region divided image; and a third stage out of boundary pixels inside each region in the region divided image.
A third step of calculating the number of pixels that are edges detected by the step of -1; a number of boundary pixels inside each region obtained in the step of 3-2; Step 3-4 of calculating the ratio of the number of pixels which are edges among the boundary pixels inside each area obtained in Step 3; If the ratio of the number of inner boundary pixels to the number of edge pixels among the inner boundary pixels of the region is larger than a preset value, the region is determined as a character region and left in the region divided image, and so on. If not, a 3-5th step of determining the area as a background noise area and deleting it from the area divided image is provided. 3. An image input storage means for inputting an image in which characters are displayed and storing it as an original image, and connecting the original images input and stored by the image input storage means using a predetermined method. Image region dividing means for dividing the image into pixel regions to obtain a region divided image; and for each region in the region divided image obtained by the image region dividing device, the contrast between the inside of the region at the region boundary and the outside of the region. A high contrast area determining unit that calculates a contrast feature representing a difference between the two, leaves an area where the contrast feature is larger than a predetermined value as a character area in the area divided image, and deletes an area that is not so from the area divided image. A character region image storage unit that stores a character region image including all connected pixel regions determined as the character region obtained by the high contrast region determination unit; Image input storage means, the image region dividing unit, and control means for controlling the execution order of the high-contrast region determination unit and the character area image storing means, a character region determining apparatus characterized by comprising. 4. The character area extracting apparatus according to claim 3, wherein the high contrast area determining means includes: an edge detecting means for detecting an edge pixel in the original image by using a predetermined method; Region perimeter calculating means for calculating the number of boundary pixels inside each region in the region, and the number of pixels which are edges detected by the edge detection device among the boundary pixels inside each region in the region-divided image. And the number of boundary pixels inside each region obtained by the region perimeter calculation unit, and the number of boundary pixels inside each region obtained by the region boundary edge calculation unit Region perimeter / edge ratio calculating means for calculating the ratio of the number of pixels which are edges of the region; and the number of boundary pixels inside the region obtained by the region perimeter / edge ratio calculating device and the boundary inside the region. If the ratio of the number of pixels that are edges among the boundary pixels is larger than a preset value, the area is determined as a character area and is left in the area divided image; otherwise, the area is used as a background noise area. A character region determining unit for determining and deleting from the region divided image. 5. A character area determining method according to claim 1, wherein a program for causing a computer to execute the steps in the character area determining method according to claim 1 or 2 is recorded on a computer-readable recording medium. Recording medium on which is recorded.