JP3480234B2 - Image processing device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention performs proper image processing according to an attribute of a document image in which images having various attributes such as a character image, a halftone image, or a grayscale image (photographic image) are mixed. The image processing apparatus that can be used is used, for example, in a digital copying machine.

[0002]

2. Description of the Related Art As an original used in a copying machine or the like, an original having only a character image, an original having only a photographic image, an original having only a halftone dot image, and an original having a mixture thereof are often used. It is desirable that the image processing for these various images should be adapted to each characteristic.

To do this, the original image is converted into characters / photos /
It is necessary to accurately determine the area to be divided into areas according to attributes such as halftone dots. In the conventional area discrimination, for each pixel in the original image, the density difference between the target pixel and its peripheral pixels, the difference between the maximum value and the minimum value in the area including the target pixel, or density information such as a histogram A method of determining the attribute of the pixel of interest based on the above is often used. Also,
There is also known a method of discriminating an attribute in a region from a frequency component in the region including the pixel of interest. Neural nets are often used to discriminate between these regions.

[0004]

However, in the above-described conventional area discriminating method, the attribute of each pixel in the original image or the attribute of a narrow area including peripheral pixels is discriminated. Misclassifications often occur in attributes.

Therefore, there is a possibility that a plurality of attribute areas, which are erroneously discriminated as different attributes, may be scattered within one actual attribute area of the original image. Then, even though the areas actually have the same attribute, a plurality of types of image processing having the content corresponding to the attribute that is erroneously discriminated is executed, which causes a problem that image quality is deteriorated due to erroneous discrimination. It was

The present invention has been made in view of the above-mentioned problems, and even when a document image includes a plurality of types of images, it is possible to accurately determine the area by extracting the boundary line of each image. An object of the present invention is to provide an image processing apparatus capable of performing appropriate image processing for each area according to an attribute.

[0007]

According to another aspect of the present invention, there is provided a device for extracting a feature amount for each pixel of an input image, and a device for determining an attribute for each pixel based on the feature amount. An attribute correction unit that applies a Hough transform to pixels located at the boundary of the attribute to detect a straight line, divides the image into a plurality of local regions by the detected straight line, and determines the attribute for each of the divided local regions. And means for performing image processing corresponding to the attribute of each of the local areas.

In the apparatus according to the invention of claim 2, the attribute is
The correction means Hough transforms the pixels located at the boundary of the attribute.
To extend the line segment extracted by applying
Divide the image into multiple local regions by lines and divide
The attribute is determined for each local area. Invention of Claim 3
The device relating to
For pixels located at the boundary of the attribute in the local area
The Hough transform is applied again to detect a straight line, and the detected straight line is
Therefore, the local area is subdivided. In the apparatus according to the invention of claim 4, the attribute correction means extracts a boundary point which is a pixel located at a boundary based on the attribute determined for each pixel, and the boundary point on the orthogonal coordinate plane. On the other hand, a means for applying the Hough transform to convert to a curve on the polar coordinate plane, a means for detecting the intersection of each curve converted to the polar coordinate plane, and the inverse conversion of the polar coordinates of the detected intersection to a straight line on the orthogonal coordinate plane. By doing so, it has a unit for generating a boundary line, a unit for extracting a region surrounded by the boundary line as a local region, and a unit for discriminating an attribute for each local region.

According to a fifth aspect of the present invention, the attribute of each of the local areas is determined based on the attribute determined for each of the pixels. The apparatus according to the invention of claim 6 has means for performing noise removal processing on the attribute discrimination result for each pixel, and performs the Hough transform on the pixel subjected to noise removal processing. Apply.

The process according to the present invention will be described with reference to FIGS. 7 and 8. When attribute determination is performed for each pixel on the original image RF conceptually shown in FIG. 7A, as shown in FIG. 7B, the pixel GA1 and the area A2 which are determined to be the area A1 are displayed.
The attribute map MP1 in which the pixels GA2 determined to be the above are mixed is obtained.

By removing noise from the attribute map MP1 shown in FIG. 7B, an attribute map MP2 shown in FIG. 7C is obtained. FIG. 8D is a boundary map MP3 showing the result of extracting the boundary point GB between the area A2 and the non-area A2 with respect to the attribute map MP2 shown in FIG. 7C.

FIG. 8E shows the result of detecting the line segments LB3 and LB4 having a predetermined length or more by applying the Hough transform to the boundary point GB. It is divided into four local regions AL1 to AL4 with the line segments LB3 and LB4 as boundaries.
The attributes are discriminated in each of these local areas AL1 to AL4.

FIG. 8F shows the result of adopting the attribute of the pixel GA having the larger number of pixels GA1 and pixels GA2 in each of the local areas AL1 to AL4. That is, the local areas AL1 and AL2 are areas A1, and the local areas AL3 and AL4 are areas A2, which is the actual area A of the original image RF.
1, close to A2.

When dividing the local area AL, the local area AL is divided into local areas AL recursively until no line segment having a predetermined length or more is detected. Is also possible.

According to the present invention, the Hough transform is applied to the region discrimination result for each pixel, the image is divided into a plurality of local regions, and the attribute is determined for each local region.
In the local area, the same attribute is given to enable uniform image processing, and image quality deterioration due to misjudgment is prevented.

[0016]

1 is a block diagram of an image processing unit 2 according to the present invention.
3 is a diagram showing an overall configuration of an image reading device 1 having a built-in device 3. FIG.

In FIG. 1, the image reading apparatus 1 includes a main body 10, a document cover 11, a document placing glass 12, an exposure lamp 13, mirrors 14a, b, c, a lens 15, a CCD sensor 16, a slider driving unit 17, and an operation. Panel section 21,
The control unit 22 and the image processing unit 23 are included.

The original document DR placed on the original document placement glass 12 is illuminated by the exposure lamp 13. Manuscript DR
The reflected light from is guided to the lens 15 by the mirrors 14a, 14b, 14c, and 14c, and the image of the document DR is formed on the CCD sensor 16.

Exposure lamp 13 and mirrors 14a, b, c
Is moved by the slider driving unit 17 at a speed according to the magnification, and the document D placed on the document placing glass 12 is moved.
Scan R over its entire surface.

The reflected light of the document DR incident on the CCD sensor 16 is converted into an image signal SP which is an electric signal by the CCD sensor 16 and is input to the image processing section 23. In the image processing unit 23, various analog signal processing, A / D conversion, and digital image processing are sequentially performed on the image signal SP. The image data DO that has undergone image processing is
It is output to the external device 24, for example, the image memory or the printer unit.

FIG. 2 is a block diagram showing the arrangement of the image processing section 23. The image processing unit 23 includes an A / D conversion unit 31, a shading correction unit 32, a LOG correction unit 33, an image storage unit 34, an area determination unit 35, and an image correction unit 36.

The image signal SP output from the CCD sensor 16 is converted into image data D1 which is a digital signal by the A / D converter 31, and the shading correction unit 32.
Then, the image is stored in the image storage unit 34 via the LOG correction unit 33. Based on the image data D3 stored in the image storage unit 34, the area determination unit 35 determines the attribute of each local area, and the attribute information DA2 is output. In the image correction unit 36, the parameters are switched according to the discrimination result in the region discrimination unit 35, and the image data D3 is subjected to the image processing corresponding to the attribute of each local region.

FIG. 3 is a block diagram showing the structure of the area discrimination section 35. The area discriminating unit 35 includes a feature amount extracting unit 41, an attribute discriminating unit 42, a calculating unit 43, an attribute storing unit 44, and a local region attribute storing unit 45.

The feature quantity extraction unit 41 extracts the feature quantity DC for each pixel of the input image data D3. A Laplacian filter, for example, is used as the feature quantity extraction unit 41. The attribute discrimination unit 42 discriminates the attribute for each pixel based on the feature amount DC and outputs the attribute information DA1. The attribute information DA1 is stored in the attribute storage unit 44.

The calculation unit 43 outputs the corrected attribute information DA2 by performing a calculation for discriminating the attribute for each local area based on the attribute information DA1. That is, the calculation unit 43
On the basis of the attribute information DA1, the Hough transform is applied to the pixels located at the boundary of the attribute to detect a straight line, the image data D3 is divided into a plurality of local regions by the detected straight line, and each of the divided local regions is divided. Determine the attributes for. The attribute for each local area is attribute information DA. The local area attribute storage unit 45 stores the attribute information DA2. Arithmetic unit 4
The attribute correction unit AM is configured by the 3 and the attribute storage unit 44.

Therefore, in the image data D3 read from the image storage unit 34, the feature amount extraction unit 41 extracts the feature amount DC for each pixel, and the attribute determination unit 42 determines the attribute for each pixel and obtains it. The attribute information DA1 is stored in the attribute storage unit 44. The attribute information DA1 stored in the attribute storage unit 44 is read by the calculation unit 43, and Hough transformation is applied to extract the local area AL and determine its attribute. The attribute information DA2 as the determination result is stored in the local area attribute storage unit 45.

FIG. 4 is a block diagram functionally showing the structure of the attribute correction unit AM. The attribute correction unit AM includes a noise removal processing unit 51, a boundary point extraction unit 52, a Hough conversion unit 53, an intersection detection unit 54, an inverse conversion unit 55, a local area extraction unit 56, and an attribute determination unit 57.

The noise removal processing section 51 uses the attribute information DA1.
Processing for removing the noise included in. The noise removal processing unit 51 deletes a candidate area of another image having a small area included in the candidate area from the candidate area of the photographic image and the halftone dot image. For example, an area area having a character size of 14 points or less, which is included in the candidate areas of the photographic image and the halftone dot image, is removed. As a result, the attribute information DA1a is obtained.

The boundary point extraction unit 52 extracts the boundary point PB which is a pixel located at the boundary based on the attribute information DA1a which is the information of the attribute determined for each pixel. The Hough transform unit 53 applies Hough transform (Hough transform) to each boundary point PB on the orthogonal coordinate plane to transform into a curve on the polar coordinate plane. The intersection detection unit 54 detects the intersection PC of each curve converted into the polar coordinate plane. The inverse conversion unit 55 generates the boundary line LB by inversely converting the detected polar coordinates of the intersection PC into a straight line on the orthogonal coordinate plane. The local area extraction unit 56 extracts the area surrounded by the boundary line LB as the local area AL.

The attribute discriminating section 57 discriminates the attribute of each local area AL and outputs the attribute information DA2 which is the discrimination result. In determining the attribute for each local area AL, the attribute determined for each pixel is used. For example, if 50% or more of pixels have a specific attribute in one local area AL, the attribute of that pixel is set as the attribute of the local area AL. When pixels of three or more attributes are mixed in the local area AL, the attribute having the largest number of pixels among them is set as the attribute of the local area AL.

FIG. 5 is a block diagram showing the structure of the image correction section 36. In FIG. 5, the image correction unit 36 displays
Movement processing unit 61, MTF correction unit 62, density correction unit 63,
And a binarization processing unit 64. In each of these units, the parameters for the respective processes are set based on the attribute information DA2 output from the area discrimination unit 35. As a result, image processing corresponding to the attribute is performed for each local area AL. For example, edge enhancement processing for character areas, smoothing processing for photographic images,
Filtering processing is performed on the halftone image.

Next, the processing operation of the area discriminating section 35, particularly the processing operation of the attribute information DA will be described in more detail.
First, the Hough transform will be briefly described. FIG. 6 is a diagram showing the concept of Hough transform. Of these, FIG. 6 (A) shows x−
The positions of the points A to G in the y space (orthogonal coordinate plane) are shown. In FIG. 6B, the points A to G are ρ-θ by Hough transform.
The curve shown on the space (polar coordinate plane) is shown.

The Hough transform is a method for detecting a figure such as a straight line, a circle or an ellipse from an image. In particular, Du
A straight line detection method by the method of da and Hart is often used, and this method is also used in this embodiment. In this method, the straight line is expressed by ρ = xcos θ + ysin θ (1). Here, ρ is the length of a perpendicular line drawn from the origin to a straight line, and θ is the angle between the perpendicular line and the x-axis.

A point on the xy space shown in FIG. 6A, for example point A (x0, y0), is expressed by the equation (1) as shown in FIG.
In the ρ-θ space shown in (B), it corresponds to the locus ρ = x0 cos θ + y0 sin θ of one sine curve.

On the contrary, the locus of one sinusoidal curve in the ρ-θ space is (x0, y0) in the xy space.
Represents all straight lines passing through. Therefore, xy
When each point on one straight line in the space is converted into the ρ-θ space, the loci of the sinusoidal curves formed on the ρ-θ space from these points intersect at one point. By inversely transforming the points at which the trajectories intersect, it becomes 1 in the xy space.
The straight line of the book is obtained. That is, according to this method, one straight line can be detected based on the discrete points, in other words, even if the straight line has a break.

In FIG. 6A, points A, B, C and D on the xy space are points on one straight line LB1. On the straight line LB1, θ is π / 4. When each of these points is transformed into the ρ-θ space by the Hough transform, as shown in FIG. 6B, the loci become sinusoidal curves, and the loci intersect at the intersection α.

Similarly, points E, B, F, G on the xy space
Is a point on one straight line LB2. On the straight line LB2, θ is π / 2. When each of these points is transformed into the ρ-θ space by the Hough transform, a locus of a sinusoidal curve as shown in FIG. 6B is obtained, and the loci intersect at the intersection point β.

Thus, a straight line parallel or perpendicular to the axis in the xy space has θ = 0, π / in the ρ-θ space.
There is an intersection at the 2 and π positions. Therefore, θ = 0, π /
If there is an intersection at the position of 2 or π, the intersection is xy
A straight line (line segment) can be obtained by inversely transforming in space.

As a concrete method for obtaining the straight line, the pixel position which is the element candidate of the straight line is detected from the original image, and each point of the locus in the ρ-θ space corresponding to the pixel position is obtained. The obtained points are counted up with respect to the coordinate position in the ρ-θ space. Since a point (maximum value point) having a count value equal to or greater than a certain threshold is an intersection point, a straight line on the xy plane is extracted by performing inverse transformation using the equation (1) on the maximum value point. It

Further, the length of the straight line (line segment) is detected based on the coordinate position on the xy plane corresponding to the locus passing through the maximum value point. For example, in FIG. 6, A, B, C,
When it is detected that each point of D is on a straight line, the end points E and G among these four points may be detected and the distance between these two points may be measured. Therefore, it is possible to extract only the line segment having a predetermined length or more.

Next, the operation of the attribute correction process in the image reading apparatus 1 will be described. 7 and 8 are diagrams for conceptually explaining the operation of the attribute correction processing. FIG. 7A shows a boundary portion between two regions A1 and A2 having different attributes in the original image RF that is the image of the document DR. The area A1 is a non-area A2, that is, a non-area A2. When the attribute determination is performed on the original image RF for each pixel, as shown in FIG. 7B, the pixel GA1 determined to be the area A1 and the pixel GA2 determined to be the area A2 are mixed. The attribute map MP1 is obtained.
Each pixel GA1 is shown in white and each pixel GA2 is shown in gray.

According to the attribute map MP1 shown in FIG. 7B, one area has entered the other area, which is a state in which noise is mixed. By removing noise from the attribute map MP1, the attribute map MP2 shown in FIG. 7C is obtained. By performing the noise removal, the number of boundary points is reduced, the calculation time is reduced, and the misjudgment is also reduced. As a method of noise removal processing, there are a method of deleting an area region having a predetermined size or less, a method of correcting an attribute map according to an attribute count value in a matrix of a predetermined size, and the like.

FIG. 8D shows a boundary point G between the area A2 and the non-area A2 with respect to the attribute map MP2 shown in FIG. 7C.
It is a boundary map MP3 showing the result of extracting B. The boundary point GB is detected for the pixel GA2, and the boundary portion of the pixel GA2 with the pixel GA1 is the boundary point GB.
Has been extracted as. Each boundary point GB is shown in black.

FIG. 8E shows the result of detecting the line segments LB3 and LB4 having a predetermined length or more by applying the Hough transform to the boundary point GB. It is divided into four local regions AL1 to AL4 with the line segments LB3 and LB4 as boundaries.
The attributes are discriminated in each of these local areas AL1 to AL4.

FIG. 8F shows the result of adopting the attribute of the pixel GA having the larger number of pixels GA1 and pixels GA2 in each of the local areas AL1 to AL4. That is, the local areas AL1 and AL2 are areas A1, and the local areas AL3 and AL4 are areas A2, which is the actual area A of the original image RF.
1, close to A2. In determining the attribute of each local area AL, the number of pixels GA1 and GA2 in each local area AL may be counted and the attribute of the pixel GA having a large count value may be adopted.

In the description of FIG. 8, the local area A
Although the division into L is performed only once, in the image reading apparatus 1, in order to reduce erroneous determination as much as possible, until the line segment having a predetermined length or more is not detected in the divided local area AL, The division into local areas AL is performed recursively.

Next, the local area AL in the attribute correction processing
The division method of will be described in detail. FIG. 9 shows a local area A
FIG. 10 is a diagram conceptually showing a method of division into L, and FIG. 10 is a diagram showing re-division within the local area AL.

The original image RF1 shown in FIG. 9A has a character image area A3 and a photographic image area A4. The Hough transform is applied to the original image RF1 as described above to draw the boundary line LB, and as shown in FIG.
It is divided into a plurality of local areas AL surrounded by B. here,
The boundary line LB is an extension of the line segment over the entire image when a line segment having a predetermined length LG or more is detected. Further, in the divided local area AL, the Hough transform is applied again in the same manner as described above, and a line segment having a predetermined length LG or more is extracted. The extracted line segment is used as the boundary line LB, and the local area AL is subdivided.

FIG. 10 (A) shows one local area AL0 shown in FIG. 9 (B). Local area AL0
Is a region surrounded by the boundary lines LB5 to LB8. By performing subdivision in the local area AL0, 2
Boundary lines LB9 and LB10 are extracted, and as a result, four local areas AL00 and A0 are extracted as shown in FIG.
It is divided into L01, AL10, and AL11.

Here, as a specific example of the length of the line segment extracted by applying the Hough transform, the predetermined length LG1 is set to, for example, 1 cm in the first division. This corresponds to about 150 dots at a density of 400 dpi. In that case, in the second division, the predetermined length LG2 is smaller than 1 cm and is 30% or more in each of the vertical and horizontal lengths of the local region AL. The predetermined length LG1 may be set to an appropriate value within the range of several mm to several cm.

The attribute of each of the divided local areas AL is determined. As is apparent from FIG. 10B, the boundary line LB extracted in the local area AL0
9 and LB10 divides the local area AL0.
Only inside. Therefore, since unnecessary division is not performed on the local area AL that does not need to be divided, the time required for calculation is shortened.

According to the above-described embodiment, even when the original image RF has a plurality of types of images such as characters / photographs / halftone dots, the Hough transform is applied to extract the boundary line LB of each image. By doing so, it is possible to accurately determine the area and perform appropriate image processing according to the attribute on each local area AL.

Next, the attribute correction process will be described with reference to the flowchart. FIG. 11 is a flowchart showing the entire attribute correction processing, FIG. 12 is a flowchart showing attribute discrimination processing by Hough transform, and FIG. 13 is a flowchart showing local area division processing.

In FIG. 11, the attribute is discriminated for each pixel (# 11), the attribute map MP is created (# 12), and the attribute discrimination processing by Hough conversion is performed (# 13). In FIG. 12, after removing noise by removing isolated points (# 21), the boundary point GB is detected (# 22). Then, local area division processing is performed (# 23), and the attribute is determined (# 24).

In FIG. 13, Hough transformation is performed (# 3
1), it recursively divides into local areas AL whose one side length is equal to or longer than a predetermined length LG (# 32). FIG. 14 is a flowchart showing another example of the attribute correction process.

In the flow chart shown in FIG. 14,
Attribute information D between the character image area and the photograph / halftone image area
The contents of the correction process of A are different. The attribute is determined for each pixel (# 41), and the attribute map MP is created for each attribute (# 4).
2). That is, the attribute map MP is created for each candidate area of each image of characters, photographs, and halftone dots.

For a candidate area of a photograph or a halftone dot (# 4
No in step 3), noise removal processing is performed for each attribute map MP (# 44). Here, the area area smaller than the maximum character size (14 points) included in the candidate area is removed. This removes most of the noise. Then, boundary point extraction processing is performed (# 45),
Local area division processing is performed (# 46), and attributes are determined (# 47).

For the candidate area of the character image (Yes in # 43), the portion overlapping the area of the photograph or halftone dot is deleted (# 48). The processing here is performed after the areas of the photograph and the halftone dot are determined in step # 47. The remaining character image area is subjected to a closing process, which is a type of morphology process (# 49). In the closing process, a disk filter corresponding to the maximum character size (14 points) is used, and the image is once thickened and then contracted.

After that, the areas of the images of characters, photographs and halftone dots are combined (# 50). Here, the pixels are combined with priority in the order in which the attribute of each pixel is accurately determined, for example, in the order of halftone dots, photographs, and characters. That is, when the areas overlap each other, the areas are secured with priority in the order of halftone dots, photographs, and characters.

The image of hair included in the photographic image is often erroneously determined to be a character image or a halftone dot image in the attribute determination for each pixel, but according to the flowchart of FIG. By applying the Hough transform, such misjudgment is corrected, and it is correctly discriminated as a photographic image. In addition, a character image may be erroneously determined to include a halftone dot image in it, but by performing a closing process on the character image and by applying the Hough transform to the halftone dot image. By performing the discrimination processing, the region is correctly discriminated.

That is, according to the flowchart of FIG. 14, first, for the original image RF, the attributes of a photograph, a halftone dot, and a character are discriminated for each pixel, and an attribute map is generated. Next, the attribute correction process by Hough transform (# 46) and the attribute correction process (# 49) by processes other than the Hough transform are performed on each of the attribute maps to obtain the corrected attribute maps. Then, the corrected attribute map is combined.

In the above embodiment, the image correction section 3
6, the configuration of the attribute correction unit AM or the local area attribute storage unit 45, the processing content, the processing order, the processing timing, and the like, the entire configuration of the image reading apparatus 1 or the configuration of each unit, etc. are appropriately changed according to the gist of the present invention can do.

[0063]

According to the inventions of claims 1 to 6 ,
Even if there are multiple types of images in the original image,
By extracting the boundary line of each image, it is possible to accurately perform the region discrimination and perform appropriate image processing according to the attribute on each region.

According to the invention of claim 5 , the total processing amount is reduced and the time required for the processing is shortened. According to the invention of claim 6 , the calculation time is reduced and the erroneous discrimination is also reduced.

[Brief description of drawings]

FIG. 1 is a diagram showing an overall configuration of an image reading apparatus including an image processing unit according to the present invention.

FIG. 2 is a block diagram showing a configuration of an image processing unit.

FIG. 3 is a block diagram showing a configuration of an area determination unit.

FIG. 4 is a block diagram functionally showing a configuration of an attribute correction unit.

FIG. 5 is a block diagram showing a configuration of an image correction unit.

FIG. 6 is a diagram showing a concept of Hough transform.

FIG. 7 is a diagram for conceptually explaining the operation of an attribute correction process.

FIG. 8 is a diagram for conceptually explaining an operation of an attribute correction process.

FIG. 9 is a diagram conceptually showing a method of dividing into local regions.

FIG. 10 is a diagram showing another division within a local area.

FIG. 11 is a flowchart showing an entire attribute correction process.

FIG. 12 is a flowchart showing an attribute discrimination process by Hough transform.

FIG. 13 is a flowchart showing local area division processing.

FIG. 14 is a flowchart illustrating another example of attribute correction processing.

[Explanation of symbols]

Reference Signs List 36 image correction unit (means for performing image processing) 41 feature amount extraction unit (means for extracting feature amount for each pixel) 42 attribute determination unit (means for determining attribute for each pixel) 51 noise removal processing unit (noise removal) 52 Boundary point extraction unit (Boundary point extraction unit) 53 Hough transform unit (Polar coordinate plane curve conversion unit) 54 Intersection point detection unit (Intersection point detection unit) 55 Inverse conversion unit ( Boundary line generating means) 56 Local area extracting section (extracting means as local area) 57 Attribute discriminating section (meaning for discriminating an attribute for each local area) AM attribute correcting section (attribute correcting means)

─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Shinji Yamamoto 1-3, 3-402 Uehara, Tamagasaki-cho, Toyohashi-shi, Aichi (56) Reference JP-A-8-202876 (JP, A) JP-A-7 -212579 (JP, A) JP-A-2-84879 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H04N 1/40-1/409 H04N 1/46 H04N 1/60

Claims (6)

(57) [Claims] 1. A unit for extracting a feature amount for each pixel of an input image, a unit for discriminating an attribute for each pixel based on the feature amount, and a Hough transform applied to a pixel located at a boundary of the attribute. Then, a straight line is detected, the image is divided into a plurality of local regions by the detected straight line, and attribute correction means for discriminating an attribute for each divided local region, and image processing corresponding to the attribute for each local region are performed. An image processing apparatus comprising: a means for performing. 2.The attribute correction means, Hough transform is applied to the pixels located at the boundary of the attribute to extract
The straight line that extends the entire line of the image
The image is divided into a plurality of local areas, and each divided local area
Determine the attributes for The image processing apparatus according to claim 1. 3.In the attribute correction means, Located at the boundary of the attribute in the divided local area
Detect the straight line by applying Hough transform again to the pixel and detect
Subdivide the local area by the straight line, The image processing apparatus according to claim 1. 4. The attribute correcting means extracts a boundary point which is a pixel located at a boundary based on the attribute discriminated for each pixel, and Hough transform is applied to each boundary point on the orthogonal coordinate plane. To convert it to a curve on the polar coordinate plane, to detect the intersection of each curve converted to the polar coordinate plane, and to reverse the polar coordinates of the detected intersection to a straight line on the orthogonal coordinate plane to demarcate the boundary line. The image processing apparatus according to claim 1, further comprising: a unit configured to generate an area, a unit configured to extract a region surrounded by the boundary line as a local region, and a unit configured to determine an attribute of each local region. The attributes of wherein each of the local region, determines based on the determined attributes for each of the pixels, the image processing apparatus according to any one of claims 1 to 4. 6. A method for performing noise removal processing on a result of discrimination of the attribute for each pixel, wherein the Hough transform is applied to pixels subjected to noise removal processing. 1 to the image processing apparatus according to claim 5.