JP2013081174A - Image processing apparatus and image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing apparatus and an image processing method.SOLUTION: An image processing apparatus includes: a binary image generation section for generating a binary image corresponding to each layer of a ternary image on the basis of the ternary image, wherein in each binary image, pixels belonging to the layer to which the binary image corresponds have a first gradation and the other pixels have a second gradation; a connected region recognition section for recognizing in each binary image second gradation connected regions disconnected from boundaries; and a character layer decision section for, if a total area ratio of the second gradation connected regions is greater than a first threshold in two binary images, and an overlap ratio of first gradation regions in the binary image that has the smaller total area of the second gradation connected regions of the two binary images to the second gradation connected regions in the binary image that has the larger total area of the second gradation connected regions is greater than a second threshold, deciding that the layer to which the binary image having the smaller total area of the second gradation connected regions of the two binary images corresponds is a character layer, and deciding that the layer to which the other binary image corresponds is a character boundary region layer.

Description

  The present invention relates to an image processing apparatus and an image processing method, and more specifically to an image processing apparatus and an image processing method for recognizing a character layer (character layer) from an image.

  To perform video indexing and video summarization, the characters in the video are very concise and reliable clues. Therefore, after detecting a caption area in a video image, OCR (optical character recognition) may be performed on the caption area to obtain caption character information. In the process, extracting characters is a very important step.

  Characters are acquired using a color clustering technique as one type of conventional technology. In this method, it is assumed that the characters have the same color. However, in some videos this hypothesis does not always hold. Also, color information is not reliable for low-quality video images. Furthermore, if the background color is close to the character color, a lot of noise is generated.

  As another type of prior art, characters are acquired using a local binarization method (for example, Niblack algorithm). However, the characteristics of this algorithm can cause a lot of noise.

  In addition, both of the above-described two methods have a problem of how to determine an actual character layer in an image after performing image processing.

  In a video image, particularly a caption area of a video image, a character having a border region (for example, a dark character having a light border region or a light character having a dark border region) is often used. In order to extract characters, the boundary region of the characters is important information. However, the conventional character extraction method does not fully utilize the feature.

  The present invention provides an image processing apparatus and an image processing method capable of recognizing a character layer (a layer of characters) in an image so as to solve at least a part of the problems of the conventional technology described above.

  One aspect of the present invention is a binary image generation unit that generates a binary image corresponding to each layer of the ternary image based on the ternary image, wherein each binary image includes the binary image. The pixels belonging to the layer corresponding to have a first gradation and the other pixels have a second gradation. A connected area recognition unit for recognizing two gradation connected areas, and a ratio of a total area of the second gradation connected areas in the two binary images is greater than a first threshold, and the two binary values In the image, the first gradation region in the binary image having a small total area of the second gradation connection region and the second gradation in the binary image having a large total area of the second gradation connection region. If the overlapping engagement rate with the connected area is larger than the second threshold value, the two binary images are not included. A character layer determining unit that determines a layer corresponding to a binary image having a small total area of the second gradation connection region as a character layer, and determines a layer corresponding to another binary image as a character boundary region layer And an image processing apparatus.

  Another aspect of the present invention is a step of generating, based on a ternary image, a binary image corresponding to each layer of the ternary image, wherein each of the binary images corresponds to the layer corresponding to the binary image. The pixel belonging to the pixel has the first gradation and the other pixels have the second gradation, and in each of the binary images, the second gradation connected region not connected to the boundary is recognized. And a ratio of the total area of the second gradation connected regions in the two binary images is larger than a first threshold, and the second gradation connected region in the two binary images The overlapping engagement ratio between the first gradation region in the binary image having a small total area and the second gradation connection region in the binary image having a large total area of the second gradation connection region is the second. If it is larger than the threshold value, the second gradation in the two binary images And determining a layer corresponding to a binary image having a small total area of the concatenation region as a character layer and determining a layer corresponding to another binary image as a character boundary region layer. .

  ADVANTAGE OF THE INVENTION According to this invention, the image processing apparatus and image processing method which can recognize a character layer in an image can be provided.

It is a block diagram which shows the structural example of the image processing apparatus which concerns on the Example of this invention. It is a figure which illustrates the image containing a character object and a character boundary area. It is a figure which shows the binary image corresponding to each layer in the ternary image produced | generated based on the image of FIG. It is a figure which shows the binary image corresponding to each layer in the ternary image produced | generated based on the image of FIG. It is a figure which shows the binary image corresponding to each layer in the ternary image produced | generated based on the image of FIG. It is a figure which shows the hole connection area | region in the binary image of FIG. 3a thru | or 3c. It is a figure which shows the hole connection area | region in the binary image of FIG. 3a thru | or 3c. It is a figure which shows the hole connection area | region in the binary image of FIG. 3a thru | or 3c. (A) thru | or (c) is a figure which shows the binary image produced by the process of the example method used for the hole connection area | region in the obtained binary image. It is a block diagram which shows the structural example of the image processing apparatus which concerns on the other Example of this invention. It is a figure which shows the outer outline of each connection area | region in the binary image shown by FIG. 3b. It is a figure which shows one character object connection area | region in FIG. 3b, and the outer outline of this connection area | region. It is a figure which shows one character object connection area | region in FIG. 3b, and the outer outline of this connection area | region. It is a figure which shows the connection area | region of one non-character object (noise) in FIG. 3b, and the outer contour of this connection area | region. It is a figure which shows the connection area | region of one non-character object (noise) in FIG. 3b, and the outer contour of this connection area | region. It is a figure which illustrates the character layer from which noise was removed. 3 is a flowchart illustrating an image processing method according to an embodiment of the present invention. 1 is a block diagram illustrating a configuration example of a computer for realizing an image processing apparatus and an image processing method according to the present invention.

  The above and other objects, features and advantages of the present invention will be apparent as shown in the detailed description of the preferred embodiments with reference to the drawings.

  Next, an embodiment for carrying out the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram illustrating a configuration example of an image processing apparatus 100 according to an embodiment of the present invention. The image processing apparatus 100 includes a binary image generation unit 110, a connected region recognition unit 120, and a character layer determination unit 130.

A ternary image is input to the image processing apparatus 100. This ternary image may be obtained by ternarizing the image using a normal method. For example, FIG. 2 is an image illustrating a caption area in a video image. In the image shown in FIG. 2, the characters are black, the background (video screen) has a progressively changing color (gradient), and there is a white border area between the characters and the background. . This image may be processed to generate a ternary diagram using known methods. For example, this image is converted into a grayscale image (the range of gradation values is 0 to 255), and the image is ternarized by comparing the gradation of each pixel with two local window thresholds (ie, The image may be divided into three layers). As a specific example, the local window threshold value may be set as T ₁ = m−k × s and T ₂ = m + k × s. Among them, m is a local (for example, 3 × 3 pixel window) average value, s is a local square variation, and the coefficient k is a value that can be set to a different value as needed. By comparing with a threshold value, a set of pixels having a gradation value smaller than T ₁ is a black color layer, a set of pixels having a gradation value larger than T ₂ is a white color layer, and a gradation value A set of pixels between T ₁ and T ₂ may be used as the intermediate layer. Further, the ternary diagram may be acquired by using other methods such as a color clustering method.

  Based on the input ternary image, the binary image generation unit 110 generates a binary image corresponding to each layer of the ternary image, that is, extracts each layer of the ternary image. In each binary image, the pixels belonging to the layer corresponding to the binary image have the first gradation, and the other pixels (background) have the second gradation.

  3a to 3c show three binary images generated by the binary image generation unit 110 based on the ternary image of the image of FIG. In the binary images of FIGS. 3a to 3c, the pixels in the corresponding layer of the ternary image are black, and the other pixels (background) are white.

  The binary image generated by the binary image generation unit 110 is supplied to the connected region recognition unit 120. The connected area recognition unit 120 recognizes a second gradation connected area that is not connected to the boundary in each binary image. Hereinafter, the second gradation connection region that is not connected to the boundary is abbreviated as a hole connection region.

  4A to 4C show hole connection regions recognized by the connection region recognition unit 120 from the binary images of FIGS. 3A to 3C.

As a specific example, the connected region recognition unit 120 may acquire a hole connected region in the binary image by the following process.
(1) Perform color reversal processing on the binary image,
(2) The first gradation connected area connected to the boundary in the binary image subjected to the color inversion processing is converted into the second gradation,
(3) The region of the first gradation in the converted binary image is set as a hole connection region.

  5a to 5c show each binary image in the exemplified processing. 5a is a binary image to be processed, FIG. 5b is a binary image obtained by color-reversing the binary image of FIG. 5a, and FIG. 5c is a binary image of FIG. 5b. It is the binary image obtained by converting the connection area | region connected to the boundary into 2nd gradation.

  In addition, this invention is not restricted to what was illustrated above, For example, you may recognize the hole connection area | region in a binary image using the analysis method of another connection area | region.

  The recognition result of the connected area of each binary image by the connected area recognition unit 120 is supplied to the character layer determination unit 130. The character layer determination unit 130 determines whether or not binary images corresponding to the character layer and the character boundary region layer exist in the three binary images based on the comparison result of the hole connection regions in the binary images. It is determined which corresponds to the character layer and which corresponds to the character boundary region layer.

  More specifically, the character layer determination unit 130 may compare the ratio of the total area (that is, the number of images) of the hole connection regions in any two binary images. When the ratio of the total area of the hole connection regions of the two binary images exceeds a predetermined threshold (for example, 20), the character layer determination unit 130 sets the layers corresponding to the two binary images as character layers and The character boundary region layer may be determined. This is because the total area of the hole connection areas in the binary image corresponding to the character boundary area layer is usually large and the total area of the hole connection areas in the binary image corresponding to the character layer is small. This is because the ratio of the total area is usually extremely high.

  When the ratio of the total area of the hole connection areas in the two binary images exceeds the threshold value, the character layer determination unit 130 uses the binary image in which the total area of the hole connection areas is small in the two binary images. The overlapping degree of overlap between the first gradation region in FIG. 5 and the hole connection region in the binary image having a large total area of the hole connection region may be further compared. When the overlapping engagement rate is larger than a predetermined threshold, the character layer determining unit 130 determines a layer corresponding to the binary image having a small total area of the hole connection regions in the two binary images as the character layer. A layer corresponding to two binary images may be determined as a character boundary region layer.

  Taking each binary image shown in FIG. 3 as an example, the character layer determination unit 130 compares the hole connection region identification results shown in FIG. 4 and compares the total area of the hole connection region shown in FIG. 4a with the hole connection shown in FIG. 4b. FIG. 3b corresponds to the case where the ratio of the total area of the area exceeds a predetermined threshold and the overlapping engagement ratio between the hole connection area shown in FIG. 4a and the black color area in FIG. 3b exceeds the predetermined threshold. The layer to be determined is determined as the character layer, and the layer corresponding to FIG. 3A is determined as the character boundary region layer.

  According to one specific embodiment of the present invention, N1 / N2 may be the overlap ratio. N1 is the number of pixels in the product set (common part) of the first gradation area in the binary image with a small total area of the hole connection area and the hole connection area in the binary image with a large total area of the hole connection area. Yes, N2 is the number of pixels in the union of the first gradation area in the binary image with a small total area of hole connection areas and the hole connection area in a binary image with a large total area of hole connection areas. Here, the overlap engagement rate may be selected from a range of 0.6 to 0.9, and may be 0.7, for example. In addition, this invention is not limited to what is illustrated here, You may measure the said overlap ratio by another concrete method.

  Further, in a certain individual case, the total area of the hole connection regions in a certain binary image is very small (for example, smaller than 1% of the total area of the binary image). Mistakes may be made. Therefore, in one embodiment, the character layer determination unit 130 may additionally employ the following determination conditions in addition to the above two determination conditions in order to improve the determination accuracy.

  When the ratio between the total area of each hole connection region and the total area of the binary image in the two binary images is larger than a predetermined threshold (for example, 0.05), the character layer determination unit 130 The character layer and the character boundary region layer are determined.

  The character layer and / or character boundary region layer determined by the image processing apparatus 100 is output as a processing result and used for post-processing, for example, character recognition. Therefore, the image processing apparatus according to the embodiment of the present invention efficiently recognizes the character layer and the character boundary area layer in the image by using the feature of the character boundary area in the image as compared with the normal image processing apparatus. Can be determined.

  In another embodiment of the present invention, the image processing apparatus may perform noise removal processing on the determined character layer.

  FIG. 6 is a block diagram illustrating a configuration example of an image processing apparatus 600 according to another embodiment of the present invention. The image processing apparatus 600 includes a binary image generation unit 610, a connected region recognition unit 620, a character layer determination unit 630, and a noise removal unit 640. The binary image generation unit 610, the connected region recognition unit 620, and the character layer determination unit 630 are similar to the binary image generation unit 110, the connection region recognition unit 120, and the character layer determination unit 130 shown in FIG. Here, the description thereof will be omitted as appropriate.

  The noise removal unit 640 performs noise removal processing on the determined character layer based on the character boundary region layer determined by the character layer determination unit 630. More specifically, the noise removing unit 640 determines, for each first gradation connected area in the binary image determined as the character layer, the character determined as the outer contour of the first gradation connected area. When the degree of coincidence with the first gradation area in the binary image in the boundary area layer is smaller than a predetermined threshold, the first gradation connection area is removed as noise. Here, the outer contour of the connection region means a contour adjacent to the outside of the connection region and the boundary of the connection region.

  Taking the binary image shown in FIG. 3 as an example, the character layer determining unit 630 determines the layer corresponding to the binary image in FIG. 3b as the character layer, and the layer corresponding to the binary image in FIG. , The noise removal unit 640 may extract the outer contour of each first gradation connected region in the binary image of the character layer. FIG. 7 shows the outer contour of each connected region in the binary image shown in FIG. 3b. Each first gradation connected region of the binary image of the character layer may include a real character object as shown in FIG. 8a and may include noise as shown in FIG. 8b. The noise removing unit 640 removes noise by comparing the outer contour of the connected area (for example, as shown in FIGS. 8b and 9b) and the first gradation area in the character boundary area layer. Since the outer contour of the connected area shown in FIG. 8b matches the area of the first gradation in FIG. 3a, the noise removing unit 640 determines that the connected area shown in FIG. 8a corresponds to the character object. Determine (determine). On the other hand, since the outer contour of the connected area shown in FIG. 9b does not match the area of the first gradation in FIG. 3a, the noise removing unit 640 removes the connected area of FIG. 9a as noise. FIG. 10 is a diagram illustrating a character layer from which noise has been removed.

  According to a specific embodiment of the present invention, N4 / N3 may be set as the degree of match. N3 is the number of pixels of the outer contour of the first gradation connected region in the binary image of the character layer, and N4 is the outer contour and character boundary region of the first gradation connected region in the binary image of the character layer This is the number of pixels in the product set with the first gradation region in the binary image of the layer. Here, the threshold value of the matching degree described above may be selected from a range of 0.85 to 0.95, for example, and may be 0.9, for example. In addition, this invention is not limited to what is illustrated here, You may measure the matching degree mentioned above with the other concrete method.

  The image processing apparatus 600 outputs the character layer from which the noise is removed by the noise removing unit 640 as a result. As a result, the image processing apparatus according to the present embodiment can provide a character layer with less noise by using the feature of the character boundary region in the image as compared with a normal image processing apparatus. The method can effectively remove noise.

  FIG. 11 is a flowchart showing an image processing method according to the embodiment of the present invention.

  In step S1110, based on the ternary image, a binary image corresponding to each layer of the ternary image is generated. In each binary image, pixels belonging to the layer to which the binary image corresponds have a first gradation, and the other pixels have a second gradation.

  In step S1120, a second gradation connected region that is not connected to the boundary is recognized in each binary image.

  In step S1130, it is determined (determined) whether the ratio of the total area of the second gradation connected regions in the two binary images is larger than a first threshold (for example, 20), and the determination result is YES. If so, the process advances to step S1140.

  In step S1140, in the binary image in which the total area of the first gradation region and the second gradation connection region in the binary image in which the total area of the second gradation connection region is small in the two binary images is large. It is determined (determined) whether or not the overlapping engagement ratio with the second gradation connection region is larger than the second threshold value. If the determination result is YES, the process proceeds to step S1150.

  In step S1150, the layer corresponding to the binary image having the small total area of the second gradation connected regions in the two binary images is determined as the character layer, and the layer corresponding to the other binary image is determined as the character boundary region. Decide with a layer.

  According to one embodiment, in step S1140, N1 / N2 is used as the overlapping engagement ratio, and N1 is the first gradation region and the second region in the binary image in which the total area of the second gradation connection region is small. Is the number of pixels of the product set with the second gradation connected area in the binary image having a large total area of the gradation connected areas, and N2 is in the binary image having a small total area of the second gradation connected areas. This is the number of pixels in the union of the second gradation connected region in the binary image in which the total area of the first gradation region and the second gradation connected region is large.

  According to one embodiment, a noise removal step is further included after step S1150. In the noise removal step, for each first gradation connected area in the binary image of the determined character layer, the outer contour of the first gradation connected area and the second image in the binary image of the character boundary area layer determined. When the degree of coincidence with the first gradation area is smaller than the third threshold value, the first gradation connection area is removed as noise.

  According to one embodiment, in the noise removal step, N4 / N3 is set as the matching degree, N3 is the number of pixels of the outer contour of the first gradation connected region in the binary image of the character layer, and N4 is This is the number of pixels in the product set of the outer contour of the first gradation connected region in the binary image of the character layer and the region of the first gradation in the binary image of the character boundary region layer.

  According to one embodiment, if the decision result in the step S1140 is YES, the ratio between the total area of each second gradation connected region and the total area of the binary image in the two binary images is a fourth value. It is determined (determined) whether or not it is larger than a threshold value (for example, 0.05). If the determination result is YES, the process proceeds to step S1150.

  It will be appreciated by those skilled in the art that the present invention may be a device, method or computer program product. Thus, the present invention may be implemented in the following specific forms, for example, complete hardware, complete software (including firmware, resident software, microcode, etc.), or software and hardware parts A combination of these may be used. Further, the present invention may use a computer software product in any tangible expression medium, and the medium includes program code usable by a computer.

  The invention may use any combination of one or more computer readable media. The computer readable medium may be a computer readable signal medium or a storage medium, and the computer readable storage medium is an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system. , Devices, parts of equipment, propagation media, or any suitable combination thereof, but is not limited to those illustrated here. Computer-readable storage media include, as a more specific example, electrical connections between one or more conductors, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasure And programmable read-only storage (EPROM or flash memory), optical fiber, compact disk (CD-ROM), optical storage, magnetic storage, or any suitable combination thereof. In this specification, a computer-readable storage medium may be any tangible medium that applies to or stores a program related to a command execution system, apparatus, or device. .

  The computer program code of the operation for carrying out the present invention may be developed in one programming language or any combination of a plurality of programming languages, for example, an object such as Java (registered trademark), Smalltalk, C ++, etc. It includes an oriented programming language, and further includes a normal procedural programming language such as a “C” programming language or similar programming language. The program code may be executed entirely on the user's computer, partially on the user's computer, as an independent software package, or partially on the user's computer. However, some may be executed by a remote computer, or may be executed completely by a remote computer or a server. In the latter case, the remote computer may be connected to the user's computer via any network, such as a regional network (LAN) or a wide area network (WAN), or (via the Internet by an Internet provider). It may be connected to an external computer.

  FIG. 12 is a block diagram showing a configuration example of a computer for realizing the image processing apparatus and the image processing method according to the present invention. In FIG. 12, the central processing unit (CPU) 1201 performs various processing based on a program stored in a read-only memory (ROM) 1202 or a program loaded from a storage unit 1208 to a random access memory (RAM) 1203. Process. The RAM 1203 may store data necessary when the CPU 1201 executes various processes as necessary.

  The CPU 1201, ROM 1202, and RAM 1203 are connected to each other via a bus 1204. An input / output interface 1205 is also connected to the bus 1204.

  The input / output interface 1205 includes an input unit 1206 (including a keyboard and a mouse), an output unit 1207 (for example, a display such as a CRT, a liquid crystal display (LCD), a speaker, etc.), and a storage unit 1208 (such as a hard disk). And a communication unit 1209 (including a network connection card such as a LAN card, a modem, etc.). The communication unit 1209 performs communication processing via a network such as the Internet.

  The drive device (drive) 1210 may be connected to the input / output interface 1205 as necessary. Further, if necessary, for example, a removable medium 1211 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory is set in the drive device 1210 so that a computer program read from the medium is installed in the storage unit 1208. May be.

  When the above-described series of processing is realized by software, a program constituting the software may be installed from a network such as the Internet or a storage medium such as a removable medium 1211.

  It should be understood by those skilled in the art that such a storage medium has a program stored therein and is removable as shown in FIG. 12, which is distributed independently of the apparatus so as to provide the program to the user. It is not limited to the medium 1211. Examples of the removable medium 1211 include a magnetic disk (including a floppy disk (registered trademark)), an optical disk (including a CD-ROM and a DVD), a magneto-optical disk (including MD (registered trademark)), and a semiconductor memory. including. Alternatively, the storage medium may be a ROM 1202, a hard disk included in the storage unit 1208, and the like, in which a program is stored, and may be distributed to a user together with a device including them.

  The above description is only a preferred embodiment of the present invention, and the scope of the present invention is not limited thereto. Any one of ordinary skill in the art based on the claims of the present invention Modifications and variations are possible, and the protection scope of the present invention is based on the claims. In addition, any embodiment of the present invention or claims does not necessarily have to realize all of the objects, advantages, or features disclosed by the present invention. Furthermore, the terms “first”, “second”, etc. referred to in this specification and the claims merely name the components or distinguish different embodiments or ranges. The upper limit or lower limit of the number of components is not limited.

  Moreover, the following additional remarks are disclosed regarding the embodiment including each of the above-described examples.

(Appendix 1)
A binary image generation unit that generates a binary image corresponding to each layer of the ternary image based on the ternary image, and in each of the binary images, pixels belonging to the layer to which the binary image corresponds A binary image generator having a first gradation and the other pixels having a second gradation;
In each of the binary images, a connected region recognition unit that recognizes a second gradation connected region that is not connected to the boundary;
The ratio of the total area of the second gradation connection area in the two binary images is larger than the first threshold value, and the total area of the second gradation connection area in the two binary images is small. The overlapping engagement ratio between the first gradation area in the binary image and the second gradation connection area in the binary image having a large total area of the second gradation connection area is larger than the second threshold. In this case, a layer corresponding to a binary image having a small total area of the second gradation connection region in the two binary images is determined as a character layer, and a layer corresponding to the other binary image is determined as a character boundary. An image processing apparatus including a region layer and a character layer determination unit that determines a region layer.

(Appendix 2)
Let N1 / N2 be the overlap engagement rate,
N1 represents the second floor in the binary image in which the total area of the first gradation area and the second gradation connection area is large in the binary image in which the total area of the second gradation connection area is small. The number of pixels in the intersection set with the key connected region,
N2 is the second floor in the binary image in which the total area of the first gradation area and the second gradation connection area is large in the binary image in which the total area of the second gradation connection area is small. The image processing apparatus according to appendix 1, which is the number of pixels in the union with the key connection area.

(Appendix 3)
For each first gradation connected area in the determined binary image of the character layer, an outer contour of the first gradation connected area and a first in the binary image of the determined character boundary area layer The image processing apparatus according to appendix 1 or 2, further including a noise removal unit that removes the first gradation connection area as noise when the degree of coincidence with the gradation area is smaller than a third threshold value.

(Appendix 4)
N4 / N3 is the degree of match,
N3 is the number of pixels of the outer contour of the first gradation connection region in the binary image of the character layer,
N4 is the number of pixels in the product set of the outer contour of the first gradation connected region in the binary image of the character layer and the region of the first gradation in the binary image of the character boundary region layer. The image processing apparatus according to appendix 3.

(Appendix 5)
When the ratio of the total area of each of the second gradation connected regions and the total area of the binary image in the two binary images is greater than a fourth threshold, the character layer determining unit determines the character layer The image processing apparatus according to appendix 1 or 2, wherein the character boundary region layer is determined.

(Appendix 6)
Generating a binary image corresponding to each layer of the ternary image based on the ternary image, wherein in each of the binary images, the pixels belonging to the layer to which the binary image corresponds correspond to a first floor; A step, the other pixel has a second gradation,
Recognizing a second gradation connected region not connected to the boundary in each of the binary images;
The ratio of the total area of the second gradation connection area in the two binary images is larger than the first threshold value, and the total area of the second gradation connection area in the two binary images is small. The overlapping engagement ratio between the first gradation area in the binary image and the second gradation connection area in the binary image having a large total area of the second gradation connection area is larger than the second threshold. In this case, a layer corresponding to a binary image having a small total area of the second gradation connection region in the two binary images is determined as a character layer, and a layer corresponding to the other binary image is determined as a character boundary. Determining an area layer.

(Appendix 7)
Let N1 / N2 be the overlap engagement rate,
N1 represents the second floor in the binary image in which the total area of the first gradation area and the second gradation connection area is large in the binary image in which the total area of the second gradation connection area is small. The number of pixels in the intersection set with the key connected region,
N2 is the second floor in the binary image in which the total area of the first gradation area and the second gradation connection area is large in the binary image in which the total area of the second gradation connection area is small. The image processing method according to appendix 6, which is the number of pixels in the union with the key connection area.

(Appendix 8)
For each first gradation connected area in the determined binary image of the character layer, an outer contour of the first gradation connected area and a first in the binary image of the determined character boundary area layer The image processing method according to appendix 6 or 7, further including a step of removing the first gradation connection area as noise when the degree of coincidence with the gradation area is smaller than a third threshold value.

(Appendix 9)
N4 / N3 is the degree of match,
N3 is the number of pixels of the outer contour of the first gradation connection region in the binary image of the character layer,
N4 is the number of pixels in the product set of the outer contour of the first gradation connected region in the binary image of the character layer and the region of the first gradation in the binary image of the character boundary region layer. The image processing method according to attachment 8.

(Appendix 10)
The character layer and the character boundary region layer when the ratio of the total area of the second gradation connection regions and the total area of the binary images in the two binary images is greater than a fourth threshold value; The image processing method according to appendix 6 or 7, wherein the determination is performed.

(Appendix 11)
The program for making a computer perform each step as described in any one of appendix 6 thru | or 10.

(Appendix 12)
A computer-readable storage medium storing the program according to appendix 11.

  The preferred embodiments and examples of the present invention have been described above, but the present invention is not limited to these embodiments and examples, and all modifications to the present invention are within the technical scope of the present invention without departing from the spirit of the present invention. Belonging to.

100, 600 Image processing device 110, 610 Binary image generation unit 120, 620 Connected region recognition unit 130, 630 Character layer determination unit 640 Noise removal unit

Claims (10)

A binary image generation unit that generates a binary image corresponding to each layer of the ternary image based on the ternary image, and in each of the binary images, pixels belonging to the layer to which the binary image corresponds A binary image generator having a first gradation and the other pixels having a second gradation;
In each of the binary images, a connected region recognition unit that recognizes a second gradation connected region that is not connected to the boundary;
The ratio of the total area of the second gradation connection area in the two binary images is larger than the first threshold value, and the total area of the second gradation connection area in the two binary images is small. The overlapping engagement ratio between the first gradation area in the binary image and the second gradation connection area in the binary image having a large total area of the second gradation connection area is larger than the second threshold. In this case, a layer corresponding to a binary image having a small total area of the second gradation connection region in the two binary images is determined as a character layer, and a layer corresponding to the other binary image is determined as a character boundary. An image processing apparatus including a region layer and a character layer determination unit that determines a region layer. Let N1 / N2 be the overlap engagement rate,
N1 represents the second floor in the binary image in which the total area of the first gradation area and the second gradation connection area is large in the binary image in which the total area of the second gradation connection area is small. The number of pixels in the intersection set with the key connected region,
N2 is the second floor in the binary image in which the total area of the first gradation area and the second gradation connection area is large in the binary image in which the total area of the second gradation connection area is small. The image processing apparatus according to claim 1, wherein the number of pixels in the union with the key connection region is set.   For each first gradation connected area in the determined binary image of the character layer, an outer contour of the first gradation connected area and a first in the binary image of the determined character boundary area layer 3. The image processing apparatus according to claim 1, further comprising a noise removal unit that removes the first gradation connection area as noise when the degree of coincidence with the gradation area is smaller than a third threshold value. 4. N4 / N3 is the degree of match,
N3 is the number of pixels of the outer contour of the first gradation connection region in the binary image of the character layer,
N4 is the number of pixels in the product set of the outer contour of the first gradation connected region in the binary image of the character layer and the region of the first gradation in the binary image of the character boundary region layer. The image processing apparatus according to claim 3.   When the ratio of the total area of each of the second gradation connected regions and the total area of the binary image in the two binary images is greater than a fourth threshold, the character layer determining unit determines the character layer The image processing apparatus according to claim 1, wherein the character boundary region layer is determined. Generating a binary image corresponding to each layer of the ternary image based on the ternary image, wherein in each of the binary images, the pixels belonging to the layer to which the binary image corresponds correspond to a first floor; A step, the other pixel has a second gradation,
Recognizing a second gradation connected region not connected to the boundary in each of the binary images;
The ratio of the total area of the second gradation connection area in the two binary images is larger than the first threshold value, and the total area of the second gradation connection area in the two binary images is small. The overlapping engagement ratio between the first gradation area in the binary image and the second gradation connection area in the binary image having a large total area of the second gradation connection area is larger than the second threshold. In this case, a layer corresponding to a binary image having a small total area of the second gradation connection region in the two binary images is determined as a character layer, and a layer corresponding to the other binary image is determined as a character boundary. Determining an area layer. Let N1 / N2 be the overlap engagement rate,
N1 represents the second floor in the binary image in which the total area of the first gradation area and the second gradation connection area is large in the binary image in which the total area of the second gradation connection area is small. The number of pixels in the intersection set with the key connected region,
N2 is the second floor in the binary image in which the total area of the first gradation area and the second gradation connection area is large in the binary image in which the total area of the second gradation connection area is small. The image processing method according to claim 6, wherein the number of pixels in the union with the key connection region is set.   For each first gradation connected area in the determined binary image of the character layer, an outer contour of the first gradation connected area and a first in the binary image of the determined character boundary area layer The image processing method according to claim 6, further comprising a step of removing the first gradation connection area as noise when the degree of coincidence with the gradation area is smaller than a third threshold value. N4 / N3 is the degree of match,
N3 is the number of pixels of the outer contour of the first gradation connection region in the binary image of the character layer,
N4 is the number of pixels in the product set of the outer contour of the first gradation connected region in the binary image of the character layer and the region of the first gradation in the binary image of the character boundary region layer. The image processing method according to claim 8.   The character layer and the character boundary region layer when the ratio of the total area of the second gradation connection regions and the total area of the binary images in the two binary images is greater than a fourth threshold value; The image processing method according to claim 6, wherein the determination is performed.

Images