JP2022113965A - Image processing apparatus - Google Patents

Abstract

To appropriately separate a foreign substance from a specific target.SOLUTION: An image processing apparatus includes: object extraction means 211 that extracts an object that connects pixels with pixel values satisfying a predetermined value for each pixel in an image; object feature quantity calculation means 212 that calculates a feature quantity relating to a color for each of the object based on the pixel value of a region other than an outer edge region of the object; classification means 213 that classifies each of the objects into either of two clusters based on the feature quantity of each of the objects; cluster feature quantity calculation means 214 that calculates a cluster feature quantity including at least either one of a color or an area for the classified two clusters; and target extraction means 215 that extracts a region of a specific target in the image based on the two cluster feature quantities.SELECTED DRAWING: Figure 2

Description

The present invention relates to an image processing apparatus, and more particularly to an image processing apparatus that properly separates a specific object from a foreign object.

In general, a captured image may include not only an image to be extracted but also an image to be removed, and it is necessary to separate these images appropriately.

FIG. 1 is an explanatory diagram illustrating image extraction using a general threshold. (a) shows a document image of a thin piece of a cyan ink dot, and FIGS. An example is shown.

As shown in FIG. 1A, the document image includes an image of the ink fixing portion area IK101 and images of the foreign matter areas IB101 and 103. As shown in FIG.

When trying to extract only the image of the ink fixation area IK101, which is a specific object, using the technique of Patent Document 1, the color pixel values (for example, saturation pixel values) of all pixels in the document image are is extracted, and a desired value (eg, 15) is set as a threshold, and pixels with pixel values of "0" to "15" are not extracted and are controlled to be displayed in white.

In this case, as shown in FIG. 1(b), images of the foreign matter areas IB101 and 103, which are obviously not the image of the ink fixing portion area IK101, are also extracted as seen by the human eye.

Also, as shown in FIG. 1(c), the threshold is changed to "42" so as not to extract the images of the foreign matter areas IB101 and 103, and pixels with pixel values of "0" to "42" are not extracted. As a result, not only the foreign matter but also the outer edge of the ink fixing portion area IK101 are not extracted, and the entire ink fixing portion area IK101 cannot be extracted accurately. This occurs because the pixel value of the outer edge of the ink fixing portion area IK101 is equal to the pixel value of the foreign matter.

As described above, it is difficult to accurately extract only the ink-fixed portion with the general image extraction technique based on the threshold value.

Here, as a technology to separate the image to be extracted from the image to be removed in the photographed image, the target is mainly the image of the dentition in the oral cavity, and the bright mass (object) reflected in the tooth is the diseased part to be diagnosed. There is disclosed a technique related to a method of outputting an image in which the affected area is easy to see, by comparing with a threshold whether the object is a noise due to diffused reflection of light, and deleting the object determined as noise (for example, Patent Document 1 ).

Japanese Patent No. 5869822

However, in the technique described in Patent Document 1, the determination means is based only on the area of the object, and if the affected area is about the same as the noise, the two cannot be accurately determined. I had to give.

As described above, in the extraction method using the threshold value, in the case of an image including an object such as an ink-fixed object and a foreign object, the pixel value of the outer edge of the ink-fixed portion is close to the pixel value of the foreign object. , could not be done accurately and effortlessly.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an image processing apparatus capable of appropriately separating a specific object from a foreign object.

In order to achieve the above object, the features of the image processing apparatus according to the present invention are:
An image processing device for extracting a specific object from an image in which a specific object and a foreign substance are mixed,
an object extracting means for extracting an object obtained by connecting pixels having a pixel value satisfying a predetermined value for each pixel in the image;
object feature amount calculation means for calculating a feature amount related to color for each object based on pixel values of areas other than the outer edge area of the object;
Classification means for classifying each object into one of two clusters based on the feature amount of each object;
Cluster feature quantity calculation means for calculating a cluster feature quantity including at least one of color and area for the two classified clusters;
object extracting means for extracting a region of the specific object from the image based on the two cluster feature amounts;
It is to prepare

According to the characteristics of the image processing apparatus according to the present invention, it is possible to appropriately separate a specific target object and a foreign substance.

(a) shows an original image of a thin piece of a cyan ink dot, and (b) to (c) show an example of an image extracted based on a threshold in the technique described in Patent Document 1. showing. 1 is a block diagram showing the hardware configuration of an image processing apparatus according to an embodiment of the present invention; FIG. FIG. 4 is a diagram showing an example of an object extracted by an object extracting unit included in the image processing apparatus according to the embodiment of the present invention; (a) is a diagram in which the objects L1 to L14 are plotted in the feature amount space, and FIG. 4(b) is a diagram in which the objects L1 to L14 plotted in the feature amount space are classified into two clusters. FIG. 4 is a diagram showing cluster feature amounts calculated by cluster feature amount calculation means included in the image processing apparatus according to the embodiment of the present invention; FIG. 4 is a diagram showing an example of an image output to an external device connected to the image processing apparatus according to the embodiment of the invention; 4 is a flow chart showing processing contents in the image processing apparatus according to the embodiment of the present invention;

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the drawings. The same or equivalent reference numerals are attached to the same or equivalent parts and components throughout each drawing. However, it should be noted that the drawings are schematic and differ from reality. In addition, there are portions with different dimensional relationships and ratios between the drawings.

Further, the embodiments shown below are examples of apparatuses and the like for embodying the technical idea of the present invention. not specific to The technical idea of this invention can be modified in various ways within the scope of claims.

FIG. 2 is a block diagram showing the hardware configuration of the image processing apparatus 10 according to the embodiment of the invention. As shown in the figure, the image processing apparatus 10 has a controller 20 and an image reading mechanism 30 . The image processing apparatus 10 removes the background of the document from the image data of the document read by the image reading mechanism 30 and outputs the data to an external device.

The image reading mechanism 30 includes a light source, a platen, a platen cover, a CCD that is an array of charge-coupled imaging elements, a drive motor that scans the CCD, an AD converter that converts the output signal of the CCD into a digital signal, and the like. Image data of the exposed surface of the original is obtained by irradiating the original placed on the original table with light from a light source and reading reflected light from the exposed surface with a CCD. The platen cover serves to hold down the document placed on the platen and to block external light. This platen cover is white in order to reduce the influence on the image data when the document is smaller than the reading range of the CCD. Also, the document platen cover may be provided with an auto sheet feeder function.

The controller 20 has a CPU 21 , a ROM 22 , a RAM 23 and an interface 24 . The CPU 21 performs processing according to a program recorded in the ROM 22, thereby controlling the optical document reading operation of the image reading mechanism 30, and extracting an image of the ink fixing portion area from the read document image appropriately. process. The RAM 23 is used as a work area at this time. The interface 24 communicates with an external device using a predetermined protocol such as USB. Here, the CPU 21 performs image processing for appropriately extracting the image of the ink fixing portion area from the read document image, but the present invention is not limited to this. For example, image processing may be performed to appropriately extract the image of the ink fixing portion area from the document image acquired via the interface 24 .

The CPU 21 functionally includes object extraction means 211 , object feature quantity calculation means 212 , classification means 213 , cluster feature quantity calculation means 214 , object extraction means 215 , and image output means 216 .

The object extracting means 211 extracts an object in which pixels having pixel values satisfying a predetermined value are connected to each pixel in the document image shown in FIG. 1(a).

FIG. 3 is a diagram showing an example of objects extracted by the object extraction means 211 included in the image processing apparatus 10 according to the embodiment of the present invention.

As shown in FIG. 3, the object extracting means 211 extracts objects L1 to L16 by connecting pixels having pixel values satisfying a predetermined value with respect to each pixel in the document image. Here, L1 to L16 are given as labels so that each object can be identified.

The object feature amount calculator 212 calculates a feature amount related to color for each of the objects L1 to L16 based on the pixel values of the areas other than the outer edge areas of the objects L1 to L16. Here, the feature amounts relating to the colors of the objects L1 to L16 are used as the respective representative colors, and the average values of R, G, and B of the pixels forming the objects L1 to L16 are calculated.

Classifying means 213 classifies each object into one of two clusters based on the feature amount of each object.

FIG. 4(a) is a diagram in which the objects L1 to L16 are plotted in the feature amount space, and FIG. 4(b) is a diagram in which the objects L1 to L16 plotted in the feature amount space are classified into two clusters. be.

As shown in FIG. 4B, the classifying means 213 classifies the objects L1 to L16 plotted in the feature amount space into one of two clusters C1 and C2 by performing cluster analysis. Here, the k-means method, which is a non-hierarchical method, is used as an algorithm for cluster analysis. There are various algorithms for cluster analysis, and any algorithm can be used. method is preferred.

As shown in FIG. 4B, objects L1-L12 are classified into cluster C1, and objects L13-L16 are classified into cluster C2.

At this stage, it is not known which of the two clusters C1 and C2 corresponds to the ink fixing portion area IK101.

Therefore, the cluster feature amount calculation means 214 calculates a cluster feature amount including at least one of color and area for the classified two clusters.

FIG. 5 is a diagram showing cluster feature amounts calculated by the cluster feature amount calculation means 214. As shown in FIG.

As shown in FIG. 5, the cluster feature amount calculator 214 obtains the area of each of the objects L1 to L12, and calculates the average of the two clusters C1 and C2 as the cluster feature amount.

Here, the average area, which is the cluster feature amount of cluster C1, is calculated as "1383.58", and the average area, which is the cluster feature amount of cluster C2, is calculated as "30.25".

The target object extracting means 215 extracts a specific target object region (ink fixing portion region IK101) from the image based on the cluster feature amounts of the two clusters C1 and C2.

As shown in FIG. 1A, the area of the object in the foreign matter areas IB101 and 103 is smaller on average than the area of the object in the ink fixing portion area IK101. Also, in the ink fixing portion area IK101, the area of the object ranges from small to large.

Therefore, the object extraction means 215 calculates the average area is extracted as the ink fixing portion area IK101. Further, the object extracting means 215 may extract the cluster C1 having a large area variation as the ink fixing portion area IK101.

As a result, the cluster C1 can be determined as the ink fixing portion area IK101, and the cluster C2 can be determined as the foreign matter areas IB101 and 103. FIG.

The image output means 216 outputs the image obtained by extracting the ink fixing portion area IK101 to an external device via the interface 24 .

FIG. 6 is a diagram showing an example of an image output to an external device.

As shown in FIG. 6, an image in which foreign matter areas IB101 and 103, which are clusters C2, are deleted from the document image and only the ink fixing portion area IK101, which is cluster C1, is extracted is output.

FIG. 7 is a flow chart showing the processing contents of the image processing apparatus 10 according to the embodiment of the present invention.

As shown in FIG. 7, when an image is input from the image reading mechanism 30 (step S101), the object extracting means 211 selects pixels having pixel values satisfying a predetermined value for each pixel in the document image. The connected objects are extracted (step S103).

In step S105, the object extraction means 211 assigns labels L1 to L16 to the extracted objects so that they can be identified.

In step S107, the object feature amount calculator 212 calculates a feature amount related to color for each of the objects L1 to L16 based on the pixel values of the areas other than the outer edge areas of the objects L1 to L16.

In step S109, the classification means 213 classifies each object into one of two clusters based on the feature amount of each object.

In step S111, the cluster feature quantity calculation unit 214 calculates a cluster feature quantity including at least one of color and area for the classified two clusters.

In step S113, the object extracting means 215 removes foreign object regions (foreign object regions IB101 and 103) from the image based on the cluster feature amounts of the two clusters C1 and C2, respectively, and extracts a specific object region ( An ink fixing portion area IK101) is extracted.

In step S115, the image output unit 216 outputs the image obtained by extracting the ink fixing area IK101 to the external device via the interface 24. FIG.

In this embodiment, the object extracting means 211 extracts an object in which pixels having pixel values satisfying a predetermined value are connected to each pixel in the document image. On the other hand, a specific region may be extracted by masking pixels corresponding to pixel values that can be obtained in a previously determined non-extraction target region as non-extraction targets.

As a result, the background area having the same color as the paper white can be masked, and the ink-fixed area can be extracted more accurately.

In the present embodiment, the cluster feature amount calculation means 214 calculates the area average of the two clusters C1 and C2 as the cluster feature amount. You may make it calculate as a feature-value.

Since there is a large color difference between the clusters, it is possible to more accurately determine which cluster is the foreign object.

(Appendix 1)
An image processing device for extracting a specific object from an image in which a specific object and a foreign substance are mixed,
an object extracting means for extracting an object obtained by connecting pixels having a pixel value satisfying a predetermined value for each pixel in the image;
object feature amount calculation means for calculating a feature amount related to color for each object based on pixel values of areas other than the outer edge area of the object;
Classification means for classifying each object into one of two clusters based on the feature amount of each object;
Cluster feature quantity calculation means for calculating a cluster feature quantity including at least one of color and area for the two classified clusters;
object extracting means for extracting a region of the specific object from the image based on the two cluster feature amounts;
An image processing device comprising:

As a result, the color representing the feature of the object is extracted as a feature amount, and based on this, the area of the specific object (ink fixation area) is extracted, so it is possible to clearly determine whether the object is a specific object or a foreign substance. Become. Therefore, by approximating the pixel values of the outer edge region of the specific object and the pixel values of the foreign matter, it is possible to accurately separate the specific object and the foreign matter. As a result, it is possible to avoid the situation in which a specific object and a foreign object cannot be accurately separated even if a determination threshold value is given in advance as troublesome prior information, as in the prior art, and the separation of the specific object and the foreign object can be avoided. Separation can be done properly.

10 image processing device 20 controller 24 interface 30 image reading mechanism 211 object extraction means 212 object feature amount calculation means 213 classification means 214 cluster feature amount calculation means 215 object extraction means 216 image output means

Claims (1)

An image processing device for extracting a specific object from an image in which a specific object and a foreign substance are mixed,
an object extracting means for extracting an object obtained by connecting pixels having a pixel value satisfying a predetermined value for each pixel in the image;
object feature amount calculation means for calculating a feature amount related to color for each object based on pixel values of areas other than the outer edge area of the object;
Classification means for classifying each object into one of two clusters based on the feature amount of each object;
Cluster feature quantity calculation means for calculating a cluster feature quantity including at least one of color and area for the two classified clusters;
object extracting means for extracting a region of the specific object from the image based on the two cluster feature amounts;
An image processing device comprising:

Images