WO2014045950A1

WO2014045950A1 - Image processing system, image processing method, and program

Info

Publication number: WO2014045950A1
Application number: PCT/JP2013/074448
Authority: WO
Inventors: 雄馬松田; 雅文矢野
Original assignee: 日本電気株式会社
Priority date: 2012-09-20
Filing date: 2013-09-04
Publication date: 2014-03-27
Also published as: JPWO2014045950A1

Abstract

In order to extract the silhouette information of an object shape by dividing a general image, an image processing system divides, from the image, the area of an object or shape included in the image. The image processing system comprises: an initial core selection unit that selects, as an initial core, out of the constituent elements constituting the image, one or more elements that are starting points to start a calculation necessary for locating the area; a peripheral element representation unit that compares each of the constituent elements with peripheral elements around that constituent element, thereby characterizing the peripheral elements as peripheral element representations; and a set generation unit that compares the characteristics of the initial core constituting the area or of the constituent elements of the area with the peripheral element representations of the peripheral elements, thereby determining whether the peripheral elements belong to the same area as the initial core, thereby generating a set of constituent elements constituting the area, thereby dividing the area from the image.

Description

Image processing system, image processing method, and program

The present invention relates to a system for recognizing or searching a general object or shape included in image information, and in particular, an image processing system for identifying a target by dividing object or shape information from image information. About.

In recent years, with the rapid spread of digital video equipment such as digital cameras, there is an increasing expectation for general object recognition that identifies what objects are included in captured images and videos. Yes.
In general object recognition, appropriate classification of image data stored without being classified in a database, retrieval of necessary image data, and the like are performed. Furthermore, general object recognition has a possibility of being applied to various uses such as extraction of a desired scene from a moving image and re-editing by cutting out only the desired scene.
Various recognition techniques, such as face recognition and fingerprint recognition, have been developed so far as techniques related to object recognition, and these are often directed to specific applications. Such a recognition technology specialized for a specific application does not operate efficiently when it is used for another application. Examples of this problem include problems such as a decrease in recognition rate, misidentification, and an increase in the amount of information processing. For this reason, development of a technique for efficiently recognizing a general object is expected.
As a method for recognizing a general object by an information processing apparatus, a method using a silhouette of an object shape has been proposed and widely used (see, for example, Patent Document 1 and Patent Document 2).
In order to recognize a general object included in an image using these methods, it is necessary to extract silhouette information from the image. However, it is not easy to extract silhouette information from a general image.
As a technique for extracting silhouette information from an image, a method of extracting silhouette information by extracting edge information by spatially differentiating luminance information in the image and extracting object contour information from the edge information is generally used. It is.
In such an edge extraction method based on differentiation processing, it is necessary to give in advance a threshold value that determines “edge” when the differential value of luminance exceeds, and “edge” when it does not exceed that value. However, in order to extract edge information that defines the outline of an object using this method, it is necessary to know a threshold value suitable for the edge information to be extracted in advance. It is difficult to do this.
On the other hand, as a technique that can avoid these problems, a technique for extracting silhouette information by clustering pixels using information belonging to pixels constituting an image has also been proposed (for example, (See Patent Document 3 and Patent Document 4).
For example, the technique disclosed in Patent Document 3 is a technique in which a band pass filter is applied to an image, an output value is ternarized by two types of threshold values, and the image is decomposed into three types of clusters. In addition, the technique disclosed in Patent Document 4 extracts edge distribution information from edge information included in an image, uses a pixel having a specific edge as an initial nucleus of clustering processing, and is located in the vicinity of the initial nucleus and is similar. This is a technique for taking in high-strength pixels (pixels whose values defined by pixel-related information such as edge information are within a threshold) into clusters. These techniques do not require pre-giving thresholds for edges.
However, in Patent Document 3, it is necessary to set a threshold value for ternarization, and in Patent Document 4, it is necessary to set a threshold value for defining “strength of similarity”. It is still difficult to do this automatically for images.

JP 10-055447 A WO2012 / 070474A1 Japanese Patent Laid-Open No. 06-076062 JP-A-10-149449

In order to recognize a general object included in an image, it is necessary to extract and specify the object shape in the form of a silhouette or the like. However, in order to extract silhouette information from a general image such as a photograph, it is necessary to perform threshold processing on information belonging to pixels constituting the image in the conventional method. It was difficult to give an applicable threshold in advance.
An object of the present invention is to provide an image processing system, an image processing method, and a program capable of extracting silhouette information of an object shape by dividing a general image such as a photograph.

One aspect of the present invention is an image processing system that divides an object or shape region included in an image from an image, and starts calculation necessary to identify the region among components constituting the image. Characterize the peripheral element as a peripheral element representation by comparing the initial kernel selection means that selects one or more elements that are the starting points to be used as the initial kernel, and each of the constituent elements and their surrounding peripheral elements Whether each peripheral element belongs to the same area as the initial nucleus by comparing the characteristics of the peripheral element representation means, the characteristics of the initial nucleus constituting the area or the constituent elements of the area, and the peripheral element representation of the peripheral element. A set generation means for determining whether or not to generate a set of constituent elements forming an area and dividing the area from an image.

According to the present invention, it is possible to provide an image processing system capable of extracting silhouette information of an object shape by dividing a general image such as a photograph.

FIG. 1 is a block diagram showing a schematic configuration of an image processing system according to the first embodiment of the present invention.
FIG. 2 is an explanatory diagram showing an example of an input image.
FIG. 3 is an explanatory diagram showing an example of an image obtained by compressing an input image.
FIG. 4 is an explanatory diagram showing an example of an initial nucleus (pixel) selected from the image of FIG.
FIG. 5 is an explanatory diagram showing an example of peripheral pixel representation from the image of FIG.
FIG. 6 is an explanatory diagram showing an example of how the set gradually expands in the image of FIG.
FIG. 7 is a block diagram showing a configuration of an initial nucleus selection unit used in the image processing system shown in FIG.
FIG. 8 is an explanatory diagram showing an example of the component feature.
FIG. 9 is a block diagram showing a configuration of a peripheral element expression unit used in the image processing system shown in FIG.
FIG. 10 is a block diagram showing the configuration of the set generation unit used in the image processing system shown in FIG.
FIG. 11 is an explanatory diagram showing an example of an element in which the difference from surrounding elements is ambiguous.
FIG. 12 is an explanatory diagram showing an example of how elements are added to the set by the element adding means.
FIG. 13 is an explanatory diagram showing an example of silhouette information generated by the image processing system shown in FIG.
FIG. 14 is a flowchart showing an example of the operation of the image processing system shown in FIG.
FIG. 15 is a block diagram showing a schematic configuration of an image processing system according to the second embodiment of the present invention.
FIG. 16 is a block diagram showing a configuration of an input image control unit used in the image processing system shown in FIG.
FIG. 17 is a flowchart composed of FIGS. 17A and 17B and showing an example of the operation of the image processing system shown in FIG.

[Summary of the Invention]
So far, two methods have been adopted to extract silhouette information from an image. One method is a method of spatially differentiating luminance information in an image, extracting a contour of an object using edge information when the differential value exceeds a threshold value. The other method is a method of clustering pixels whose luminance difference values are within a threshold as silhouette information of the same object.
However, in any of the methods, it is necessary to set in advance a threshold having a different optimum value depending on an image, and it is difficult to automatically perform this for an image handled for the first time. In other words, a method for extracting silhouette information from an image handled for the first time has not been devised so far.
In response to these problems, the inventors propose a system that performs image processing as follows.
The processing operation of this image processing system can be shown by the following three steps.
In the first step, a pixel that is a nucleus of silhouette information of an object shape included in an image, that is, an initial nucleus is selected. For example, when an image as shown in FIG. 3 composed of pixels is given to the image processing system, one (or a plurality of) pixels are selected as the “initial nucleus” as shown in FIG.
In the second step, the surrounding pixel representation is generated by comparing the color information, ie, RGB values, of each pixel and the surrounding pixel. For example, as shown in FIG. 5, a pixel within 3 pixels in each of the vertical and horizontal directions with respect to a certain pixel (center pixel) is defined as a peripheral pixel, and a difference between RGB values from the center pixel (RGB value is expressed by a three-dimensional vector. (Expressed in norm). This peripheral pixel representation is performed for all pixels. That is, all the pixels recognize a difference from surrounding pixels as an expression.
In the third step, by comparing the initial kernel and the “peripheral pixel representation” of its surrounding pixels, it is determined whether the corresponding peripheral pixel should be included in the same set as the initial kernel and included in the set If so, the same decision is made for the surrounding pixels. For example, pixels having a small difference from the initial nucleus are taken into the set, and a set that forms a silhouette of the object shape is formed. An example of how the set is gradually expanded in this way is shown in FIG.
The image processing system of the present invention that performs each step will be described below with reference to FIGS. 1 to 17A and 17B.
By adopting this embodiment, it is possible to provide an information processing system that performs image processing that makes it possible to extract silhouette information of an object shape from an image.
[First Embodiment]
[Description of configuration]
FIG. 1 is a block diagram showing a configuration of an image recognition system according to the first embodiment of the present invention. In addition, about the structure with little connection with this invention, description is simplified or abbreviate | omitted.
The illustrated image processing system includes a control unit 10 that manages the entire information processing, an image information acquisition unit 20 that acquires image data to be processed, an image information storage unit 30 that stores the acquired image data, and the like. It includes a set information storage unit 40 that stores silhouettes of objects for which image processing has been completed as a set of in-image elements (pixels), and a processing result output unit 50. The illustrated image processing system includes an initial nucleus selection unit 101, a peripheral element expression unit 102, and a set generation unit 103.
The control unit 10 manages the overall operation of information processing related to image processing.
The image information acquisition unit 20 captures image data designated by the user, such as a moving image or a photograph, into the system and stores it in the image information storage unit 30. As the capturing method, the image information designated by the user may be acquired as it is, or may be acquired after performing preprocessing such as compression by mosaic processing in order to omit the calculation amount.
Examples of image information acquired by the image information acquisition unit 20 (examples of red apples) are illustrated in FIGS. 2 and 3. FIG. 2 shows an example of the captured image itself. FIG. 3 shows an example of a diagram compressed by mosaic processing. It is also possible to automatically collect image information from a moving image or the like at an arbitrary interval or the like.
The image information storage unit 30 stores acquired image data and results obtained by processing, and intermediate data (such as initial core information and peripheral element expressions) as necessary. The image information storage unit 30 may be configured by a memory, a hard disk drive (HDD), or any storage unit.
The set information storage unit 40 stores the silhouette of an object for which image processing has been completed as a set of elements (pixels) in the image. That is, the set information storage unit 40 stores set information. Note that an external database may be referred to as the collective information storage unit 40.
The processing result output unit 50 outputs the image processing result obtained by the system. For example, the processing result output unit 50 outputs the target object or shape of the processing result recorded in the collective information storage unit 40 to a monitor or the like. Note that any information may be output as the output form.
As shown in FIG. 4, the initial nucleus selection unit 101 selects a pixel that is a nucleus of silhouette information of an object shape included in an image, that is, an initial nucleus. Specifically, the initial nucleus selection unit 101 includes three components.
That is, as shown in FIG. 7, the initial kernel selection unit 101 includes a component element characterization unit 101-1 that expresses each component (for example, pixel) constituting a given image as an element feature, It comprises an element feature comparison unit 101-2 that compares element features of constituent elements, and an optimum element selection unit 101-3 that selects an optimum element from the comparison result of the element features as the initial nucleus.
The component element characterization unit 101-1 may be any element characterization, but it is desirable to use color information in the image. For example, the component element characterization unit 101-1 uses RGB values representing color information as element features.
At this time, if the color information of the object to be watched is determined in advance, the constituent element characterization unit 101-1 uses the color information (RGB values represented by vectors) and the element color as the element features. It is also possible to use a difference (vector norm) with information. Further, if it is desired to use a conspicuous element as an initial nucleus, the component element characterization unit 101-1 uses a difference from a peripheral element of the target element as an element feature, for example, a color information difference (RGB value as a vector). It is also possible to use the norm of the expression. Furthermore, if it is desired to keep an eye on the order from the largest possible region, the component characterization unit 101-1 determines that the difference in color information is within a predetermined threshold value from surrounding elements (or elements of the entire image). It is also possible to count the number of objects, that is, the set element candidates, and express the number as a feature of the element. Further, the difference in color information is not necessarily the norm of the vector representation of RGB values, and for example, the maximum difference among the differences between RGB may be used.
An example of the element feature expressed in this way is shown in FIG. FIG. 8 shows the result of calculating the difference from the surrounding pixels for each pixel in the same manner as in FIG. 5 and then selecting the maximum value from the differences from the surrounding pixels and visualizing it. It is. That is, a larger difference from the surrounding pixels is indicated by a larger value. When a method equivalent to this example is used, it is possible to select a conspicuous pixel having a larger difference from the surrounding area.
The element feature comparison unit 101-2 compares each element characterized by the component element characterization unit 101-1.
The optimum element selection unit 101-3 selects an optimum one from those compared by the element feature comparison unit 101-2. For example, if the color information of the object to be watched is determined in advance, the optimum element selection unit 101-3 selects the element having the feature closest to the color information. The optimum element selection unit 101-3 selects an element having a maximum difference feature value with respect to the surrounding area when it is desired to use a conspicuous element as an initial nucleus. The optimal element selection unit 101-3 selects an element that maximizes the number of set element candidates when it is desired to keep an eye on an area as large as possible. The optimum element selection unit 101-3 can adopt other methods as long as it is a means that enables selection of an optimum element to be an initial nucleus.
In the example of FIG. 8, the optimum element selection unit 101-3 can select a conspicuous pixel having a larger difference from the surrounding area by selecting the largest value in FIG.
The peripheral element expression unit 102 generates a peripheral pixel expression by comparing color information, that is, RGB values, of each pixel and its peripheral pixels. Specifically, the peripheral element representation means 102 is composed of two components.
That is, as shown in FIG. 9, the peripheral element expression unit 102 specifies, for each constituent element (for example, pixel) constituting the given image, a peripheral element region specifying unit that specifies an element to be the peripheral element. 102-1 and a peripheral element characterization unit 102-2 that characterizes each peripheral element by comparing the target element and its peripheral elements.
The peripheral element region specifying unit 102-1 specifies, for each element constituting the image, an element that should be the peripheral element. In this operation, when forming a set starting from the initial nucleus, the core elements in the set including the initial nucleus are elements that are candidates for the set element when surrounding elements are taken into the set. This is done for the purpose of specifying the search range. That is, if the range of elements to be newly incorporated into the set is wide, it can be widened, and if it is narrow, it can be narrowed. It can be an element adjacent to the target element, or it can be the entire image. Good. When the range of peripheral elements is narrower, the calculation time for determining a set per time (that is, means performed in the set generation unit) can be shortened. On the other hand, elements with the same characteristics can be included in the set with higher accuracy when the range is wider.
In the example of FIG. 5, the peripheral element specifying unit 102-1 specifies as a peripheral element region an area included within 3 pixels vertically and horizontally from a certain pixel.
The peripheral element characterization unit 102-2 characterizes the peripheral element in the region defined by the peripheral element region specifying unit 102-1 by comparing it with a target element. For example, the peripheral element characterization unit 102-2 performs this by expressing the difference of the color information with the norm of the RGB value vector representation. In addition, the peripheral element characterization unit 102-2 may evaluate the difference value of the color information by binarizing depending on whether or not the threshold value is exceeded, or may ternary using two types of thresholds. By binarization or ternarization, it is possible to express a set centered on a certain element (described here as “figure”) and a set other than that (described here as “ground”). (By ternarization, it is possible to set an intermediate area that is neither area). This threshold value can be normalized by the distribution information of the entire image of the difference information of the color information, and can be generalized to a method that does not target a specific image. FIG. 5 shows an example in which peripheral pixels are characterized (represented by color information difference) around a certain pixel by this method.
The set generation unit 103 determines whether or not the corresponding peripheral pixel should be included in the same set as the initial nucleus by comparing the initial nucleus and the “peripheral pixel representation” of the surrounding pixels. If the corresponding peripheral pixel is included in the same set as the initial nucleus, the set generation unit 103 performs the same decision making for the peripheral pixel. Specifically, the set generation unit 103 includes three components.
That is, as shown in FIG. 10, the set generation unit 103 includes a set expression generation unit 103-1 that represents the characteristics of the set generated with the initial nucleus as a starting point, and a set expression and target of the target set. Each external element should be added to the target set according to the comparison result of the inter-element-set comparison unit 103-2 that compares the peripheral element representations of the external elements not included in the set to be compared and the inter-element-set comparison means And an element adding unit 103-3 for adding if it is to be added.
The set expression generation unit 103-1 represents the characteristics of the set generated with the initial nucleus as a starting point. The characteristics of the set can also be expressed by all elements added as elements of the set by the initial kernel and the element adding unit 103-3. For example, an element having color information close to the elements included in the set can be included in the set by setting the set expression to the RGB value of the initial nucleus or the average value of the RGB values of the elements included in the set. is there. However, instead of simply using only the elements included in the set as described above, it is also possible to add to the expression elements that each element of the set is “ground”. In other words, if the elements included in the set are “figure” and the elements whose color information is separated from them are “ground”, the set can be expressed by the “figure” and “ground” of the set. Become.
By using this technique and referring to the information of “ground”, for example, whether an element with an ambiguous difference from surrounding elements (pixels) as shown in a white frame in FIG. 11 is added to the set. When determining whether or not, it becomes clear how much to continue adding, and it is possible to avoid the process of finally adding the peripheral elements and finally using the entire image as an element. Therefore, it is more desirable to express using both information of "Figure" and "Ground". The means that can be adopted as the set expression generation means is not limited to this, and any means that can express a set may be used.
The inter-element-set comparison unit 103-2 compares the set expression of the target set with the peripheral element expressions of external elements not included in the target set. For example, an element that is neither a “diagram” nor a “ground” of a set is represented by a matrix of 1, -1, 0 (each element of the matrix is, for example, each element of an image), and elements outside the set (peripheral elements) , Let us say that the elements that are neither the “figure” nor the “ground” in the vicinity are represented by a matrix of 1, -1, 0, respectively. In this case, the inter-element-set comparison unit 103-2 includes an element outside the target set in the set based on an evaluation value obtained by summing all elements of the set expression and the peripheral element expression with all elements. It can be determined whether or not it should be.
That is, the inter-element-set comparison unit 103-2 determines that the element should be included in the set if the evaluation value is positive and has a large absolute value, and if the evaluation value is negative and has a large absolute value. , It can be determined that the element should not be included in the set. Further, the inter-element-set comparison unit 103-2 performs evaluation with higher accuracy by separately evaluating the evaluation value when the set value is 1 and the evaluation value when the set value is -1. It becomes possible.
The inter-element-set comparison unit 103-2 does not necessarily need to be the component shown here, and any method can be used as long as it can be determined whether an element outside the set is included in the set. Is also possible.
The element addition unit 103-3 determines whether or not each external element should be added to the target set according to the comparison result of the inter-element set comparison unit 103-2. I do. For example, the element addition unit 103-3 performs element addition when the evaluation value obtained by summing up the integration of the matrix elements exceeds a prescribed value, for example, 0. This additional element processing is desirably performed not only on external elements but also on elements in the set. It is desirable to perform element addition processing each time a new external element is added to the set. However, this may be done after several external elements are added together in order to prevent an increase in computation time. Is possible.
The element adding unit 103-3 is desirably performed until convergence to a state where there are no more elements to be added, but in order to prevent an increase in calculation time, conditions such as an upper limit of the number of calculations can be provided. FIG. 12 shows how this addition process is performed gradually, and FIG. 13 shows the state of the set when the addition is completed.
[Description of operation]
Next, an operation example of the first embodiment will be described. FIG. 14 is a flowchart illustrating an example of the operation of the image processing system according to the first embodiment.
First, the image information acquisition unit 20 acquires target image data designated by the user and records it in the image information storage unit 30 (S1001). The acquisition of image information is not limited to the one specified by the user, and the system may acquire it automatically or semi-automatically.
Next, the initial nucleus selection unit 101 selects a pixel that is a nucleus of silhouette information of an object shape included in an image, that is, an initial nucleus, by the following three steps. First, the component characterization unit 101-1 of the initial kernel selection unit 101 performs a component characterization step (S <b> 1002) for expressing each component (for example, pixel) constituting the given image as an element feature. . Next, the element feature comparison unit 101-2 of the initial core selection unit 101 performs an element feature comparison step (S1003) for comparing the element features of the respective constituent elements. Finally, the optimum element selection unit 101-3 of the initial nucleus selection unit 101 performs an optimum element selection step (S1004) for selecting an optimum element as the initial nucleus from the comparison result of the element features.
Next, the peripheral element representation unit 102 generates a peripheral pixel representation by comparing each pixel with color information, that is, an RGB value, according to the following two steps. First, the peripheral element region specifying unit 102-1 of the peripheral element expression unit 102 specifies, for each constituent element (for example, pixel) constituting the given image, an element that should be the peripheral element. The specific step (S1005) is performed. Next, the peripheral element characterization unit 102-2 of the peripheral element expression unit 102 performs a peripheral element characterization step (S1006) of characterizing each peripheral element by comparing the target element with the peripheral element. carry out.
Next, the set generation unit 103 compares the “peripheral pixel representation” of the initial nucleus and its surrounding pixels through the following three steps, so that the corresponding peripheral pixel should be included in the same set as the initial nucleus. If it is included in the set, the same decision is made for the surrounding pixels. First, the set expression generation unit 103-1 of the set generation unit 103 performs a set expression generation step (S <b> 1007) for expressing the characteristics of the set generated with the initial nucleus as a starting point. Next, the inter-element set comparison unit 103-2 of the set generation unit 103 compares the set representation of the target set with the peripheral element representations of external elements not included in the target set. Step (S1008) is performed. Next, the element addition unit 103-3 of the set generation unit 103 determines whether or not each external element should be added to the target set according to the comparison result of the inter-element set comparison unit 103-2. If it is to be added, an element addition step (S1009) for performing addition is performed. The element adding step is continued until the end condition defined by the element adding means is satisfied.
Finally, the processing result output unit 50 outputs the set generated as a result of the processing as region information that forms a silhouette of the object shape (S1010).
By operating the image processing system in this way, it can be expected to perform image processing that makes it possible to extract silhouette information of an object shape from an image.
Note that each unit of the image processing system may be realized using a combination of hardware and software. In a form in which hardware and software are combined, an image processing program is developed in the RAM, and hardware such as a control unit (CPU) is operated based on the program, thereby realizing each unit as various means. Further, the program may be recorded on a recording medium and distributed. The program recorded on the recording medium is read into a memory via a wired, wireless, or recording medium itself, and operates a control unit or the like. Examples of the recording medium include an optical disk, a magnetic disk, a semiconductor memory device, and a hard disk.
To describe the above-described embodiment in another expression, an information processing apparatus that operates as an image processing system is based on an image processing program expanded in a RAM, an initial kernel selection unit 101, a peripheral element expression unit 102, a set generation unit It can be realized by operating the control unit 10 as 103.
As described above, according to the first embodiment of the present invention, it is possible to provide image processing that makes it possible to extract silhouette information of an object shape from an image.
In addition, the specific configuration of the present invention is not limited to the above-described embodiment, and changes within a range not departing from the gist of the present invention are included in the present invention.
[Second Embodiment]
A second embodiment for carrying out the present invention will be described in detail with reference to the drawings.
[Description of configuration]
FIG. 15 is a block diagram showing a configuration of an image processing system according to the second embodiment of the present invention. In addition, about the structure with little connection with this invention, description is simplified or abbreviate | omitted.
The second embodiment assumes a case where the image size acquired in the first embodiment is very large and the calculation amount becomes enormous. In such a case, the first embodiment subdivides the acquired image, generates partial information of the silhouette once in the subdivided image, and then generates silhouette information generated in all of the subdivided images. By using the technique applied in the first embodiment again as a factor, a technique for calculating the entire silhouette information is adopted.
For this reason, the second embodiment is different from the first embodiment in that an input image control unit 201 is newly provided. In addition, some elements have functions different from those of the first embodiment.
The input image control unit 201 subdivides the acquired image and generates the partial information of the silhouette once in the subdivided image (subdivided image). In addition to performing the control for giving the inside of the image, after the input of all the subdivided images is completed, the image information having the silhouette information constituted in each subdivided image as a constituent element is given as an input. Specifically, the input image control includes three components.
That is, as shown in FIG. 16, the input image control unit 201 includes an image segmentation unit 201-1 that subdivides an image (original image) acquired by the system in advance by some method including grid division, A subdivided image input unit 201-2 that gives each of the subdivided images as an input image to the system, and an integration that gives, as an input to the system, an image having an area configured in each subdivided image as a constituent element And an image input unit 201-3.
The image subdivision unit 201-1 may subdivide the subdivision image by the method of giving the vertical and horizontal sizes in advance, or may give the number of subdivision images in advance and subdivide the image by a suitable image size. In addition, the image segmentation unit 201-may be any component as long as it is a component intended for image segmentation.
The segmented image input unit 201-2 inputs the images segmented by the image segmentation unit 201-1 one by one.
The integrated image input unit 201-3 configures an image having regions configured in the respective subdivided images as constituent elements, and performs input to the system.
[Description of operation]
Next, an operation example of the second embodiment will be described. FIG. 17 is a flowchart illustrating an example of the operation of the image processing system according to the second embodiment. In addition, about the structure with little connection with this invention, description is simplified or abbreviate | omitted.
The input image control unit 201 subdivides the acquired image by the following three steps and inputs the system in order to generate partial information of the silhouette once in the subdivided image (subdivided image). As an image, control is given to each subdivided image, and after input of all the subdivided images is completed, image information including silhouette information composed in each subdivided image as input is input. give.
First, the image segmentation unit 201-1 of the input image control unit 201 performs an image segmentation step (S2001) in which an image (original image) acquired by the system is segmented in advance by some method such as grid division. carry out. Next, the segmented image input unit 201-2 of the input image control unit 201 performs a segmented image input step (S2002) in which each segmented image is provided as an input image to the system. Finally, the integrated image input unit 201-3 of the input image control unit 201 provides an integrated image input step as an input to the system, with an image having regions configured in the respective subdivided images as components (S2003). To implement.
In each of the above-described embodiments, the processing of the embodiment includes information stored in a computer-readable storage medium encoded with a program, software, or an instruction that can be executed by a computer. It may be executed by installing it on a computer. The storage medium includes not only a portable recording medium such as an optical disk, a floppy (registered trademark) disk, and a hard disk, but also a transmission medium that temporarily records and holds data such as a network.
Although the present invention has been described with reference to the embodiments, the present invention is not limited to the above embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.
A part or all of the above-described embodiment can be described as in the following supplementary notes, but is not limited thereto.
(Supplementary Note 1) An image processing system that divides a region of an object or a shape included in an image from the image,
An initial nucleus selecting means for selecting one or more elements, which are starting points for starting calculation necessary for specifying the region, among the components constituting the image, as an initial nucleus;
A peripheral element expression means for characterizing the peripheral element as a peripheral element expression by comparing each of the constituent elements with the peripheral elements around them;
Whether each of the peripheral elements belongs to the same region as the initial nucleus by comparing the characteristics of the constituent elements of the initial nucleus or the region constituting the region and the peripheral element representation of the peripheral element. Set generation means for generating the set of the constituent elements forming the region and dividing the region from the image;
An image processing system comprising:
(Supplementary Note 2) The initial nucleus selection means includes:
Component characterization means for expressing each of the components constituting the image as element features;
Element feature comparison means for comparing the element features;
Optimal element selection means for selecting an optimal element as the initial nucleus from the comparison result of the element characteristics;
The image processing system according to claim 1, further comprising:
(Additional remark 3) The said component element characterization means is the largest among the difference of the feature-value with respect to the element contained in the defined range regarding all the said components which comprise the said element and the said image. And ii) the number of elements within which a difference is within a predetermined numerical value, and iii) the difference between the predetermined characteristic and the corresponding element characteristic, The image processing system according to appendix 2.
(Supplementary Note 4) The peripheral element expression means includes:
Peripheral element region specifying means for specifying another constituent element to be a peripheral element for each of the constituent elements constituting the image;
A peripheral element characterizing means for characterizing each peripheral element by comparing each of the constituent elements with the peripheral element;
The image processing system according to any one of appendices 1 to 3, further comprising:
(Supplementary Note 5) The peripheral element characterization unit is configured to use a difference value of a feature amount obtained by comparing the corresponding element with a peripheral element determined by the peripheral element region specifying unit, or the difference value is obtained by using two types of threshold values. The image processing system according to appendix 4, wherein the characterization is performed using the ternarized values in (4).
(Supplementary Note 6) The threshold for performing ternarization in the peripheral element characterization means is:
The supplementary note 5 is characterized in that the determination is performed by performing a normalization process using a feature amount belonging to the image, including distribution information of a difference value of a feature amount of the constituent elements constituting the image. Image processing system.
(Supplementary Note 7) The set generation means includes:
A set expression generation means for generating a set expression by using a component of the set including the initial nucleus, and a feature of the set generated using the initial nucleus as a starting point;
A comparison means between element sets for comparing the set representation of the set with the peripheral element representation of external elements not included in the set;
An element addition unit that determines whether or not the external element should be added to the set according to a comparison result of the inter-element set comparison unit;
The image processing system according to any one of appendices 4 to 6, further comprising:
(Supplementary Note 8) The set expression generation means sets the possibility of being included in all of the constituent elements constituting the image or the elements included in a predetermined range as the constituent elements of the set, or Item 8. The image processing system according to appendix 7, wherein the image processing system is expressed by a multivalued value including ternary value.
(Supplementary note 9) The inter-element-set comparison means includes:
The features in the elements included in all or a predetermined range of the constituent elements constituting the image expressed by the set expression generating means and the peripheral element characterizing means are integrated in the respective elements,
By summing up the integration results for all applicable elements,
The image processing system according to appendix 7 or 8, wherein the element is compared with the set.
(Supplementary Note 10) The image processing system includes:
Image segmentation means for segmenting the image in advance;
Subdivided image input means for giving each of the subdivided images as input images of the initial nucleus selection means, the peripheral element representation means, and the set generation means,
An integrated image input means for giving an image having a region formed in the subdivided image as a component, as an input of each of the initial nucleus selection means, the peripheral element representation means, and the set generation means;
The image processing system according to any one of appendices 1 to 9, further comprising:
(Supplementary note 11) An image processing method for dividing an object or a region of a shape included in an image from the image,
An initial nucleus selection step of selecting one or more elements, which are starting points for starting calculation necessary for specifying the region, among the components constituting the image as an initial nucleus;
A peripheral element expression step for characterizing the peripheral element as a peripheral element expression by comparing each of the constituent elements with the peripheral elements around them,
Whether each of the peripheral elements belongs to the same region as the initial nucleus by comparing the characteristics of the constituent elements of the initial nucleus or the region constituting the region and the peripheral element representation of the peripheral element. A set generation step of generating a set of the constituent elements forming the region and dividing the region from the image;
An image processing method comprising:
(Supplementary Note 12) The initial nucleus selection step includes:
A component characterization step of expressing each of the components constituting the image as element features;
An element feature comparison step for comparing the element features;
An optimal element selection step of selecting an optimal element as the initial nucleus from the comparison result of the element characteristics;
The image processing method according to appendix 11, characterized by comprising:
(Additional remark 13) The said component element characterization process is the largest of the difference of the feature-value with respect to the element included in the defined range regarding the said element and all the said component which comprises the said image. And ii) the number of elements within which a difference is within a predetermined numerical value, and iii) the difference between the predetermined characteristic and the corresponding element characteristic, The image processing method according to appendix 12.
(Supplementary Note 14) The peripheral element expression step includes:
For each of the constituent elements constituting the image, a peripheral element region specifying step for specifying another constituent element to be the peripheral element;
A peripheral element characterization process for characterizing each peripheral element by comparing each of the constituent elements with the peripheral element;
14. The image processing method according to any one of appendices 11 to 13, characterized by comprising:
(Supplementary Note 15) In the peripheral element characterization step, the difference value of the feature amount obtained by comparison between the corresponding element and the peripheral element determined by the peripheral element region specifying step, or the difference value is represented by two types of threshold values. 15. The image processing method according to appendix 14, wherein the characterization is performed using the ternary values obtained in step 14.
(Supplementary Note 16) The threshold for performing ternarization in the peripheral element characterization step is:
Item 15. The supplementary note 15, wherein the determination is performed by performing a normalization process using feature amounts belonging to the image including distribution information of difference values of feature amounts of the constituent elements constituting the image. Image processing method.
(Supplementary Note 17) The set generation step includes:
A set expression generation step of generating a set expression by using a component of the set including the initial nucleus as a feature of the set generated using the initial nucleus as a starting point;
An inter-element-set comparison step of comparing the set representation of the set with the peripheral element representation of external elements not included in the set;
An element addition step of determining whether or not the external element should be added to the set according to a comparison result of the inter-element set comparison step;
The image processing method according to claim 14, comprising:
(Supplementary Note 18) In the set expression generation step, all of the constituent elements constituting the image or elements included in a predetermined range are set to continuous values or the possibility of being included in the constituent elements of the set. 18. The image processing method according to appendix 17, wherein the image processing method is expressed by a multi-valued value including ternarization.
(Supplementary note 19) The element set comparison step includes:
The features in the elements included in all or all of the constituent elements constituting the image expressed by the set expression generation step and the peripheral element characterization step are integrated in the respective elements,
By summing up the integration results for all applicable elements,
The image processing method according to appendix 17 or 18, wherein the element is compared with the set.
(Supplementary Note 20) The image processing method includes:
An image segmentation step for segmenting the image in advance;
A subdivided image input step of giving each of the subdivided images as input images of the initial nucleus selection step, the peripheral element representation step, and the set generation step;
An integrated image input step of giving an image having an area configured in the subdivided image as a component as an input of each of the initial nucleus selection step, the peripheral element representation step, and the set generation step;
The image processing method according to any one of appendices 11 to 19, further comprising:
(Supplementary note 21) An image processing program for causing a computer to divide a region of an object or a shape included in an image from the image, wherein the computer
An initial kernel selection procedure for selecting, as an initial kernel, one or more elements that are starting points for starting calculation necessary to identify the region from among the components constituting the image;
A peripheral element expression procedure for characterizing the peripheral element as a peripheral element expression by comparing each of the constituent elements with their peripheral peripheral elements;
Whether each of the peripheral elements belongs to the same region as the initial nucleus by comparing the characteristics of the constituent elements of the initial nucleus or the region constituting the region and the peripheral element representation of the peripheral element. A set generation procedure for generating a set of the constituent elements forming the region and dividing the region from the image;
An image processing program for executing
(Supplementary Note 22) The initial nucleus selection procedure is performed by the computer.
A component characterization procedure for expressing each of the components constituting the image as element features;
An element feature comparison procedure for comparing the element features;
An optimal element selection procedure for selecting an optimal element as the initial nucleus from the comparison result of the element characteristics;
Item 22. The image processing program according to appendix 21, wherein the image processing program is executed.
(Additional remark 23) The said component element characterization procedure makes the said computer calculate the difference of the feature-value with respect to the element included in the defined range about the said element and all the said component which comprises the said image. Characterizing is performed by any one of the maximum value among them, ii) the number of elements whose difference is within a predetermined numerical value, and iii) the difference between the predetermined characteristic and the corresponding element characteristic. The image processing program according to attachment 22, wherein the image processing program is executed.
(Supplementary Note 24) The peripheral element expression procedure is performed on the computer.
For each of the constituent elements constituting the image, a peripheral element region specifying procedure for specifying another constituent element to be the peripheral element;
A peripheral element characterization procedure for characterizing each peripheral element by comparing each of the constituent elements with the peripheral element;
The image processing program according to any one of appendices 21 to 23, wherein:
(Supplementary Note 25) In the peripheral element characterization procedure, the difference value of the feature amount obtained by comparing the corresponding element with the peripheral element determined by the peripheral element region specifying procedure, or the difference value is stored in the computer. 25. The image processing program according to appendix 24, wherein characterization is performed using values ternarized with two types of threshold values.
(Supplementary Note 26) The threshold for performing ternarization in the peripheral element characterization procedure is:
25. The supplementary note 25, wherein the determination is performed by performing a normalization process using feature amounts belonging to the image including distribution information of difference values of feature amounts of the constituent elements constituting the image. Image processing program.
(Supplementary note 27) The set generation procedure is performed by the computer.
A set expression generation procedure for generating a set expression by using a component of the set including the initial nucleus as a feature of the set generated using the initial nucleus as a starting point;
An inter-element-set comparison procedure for comparing the set representation of the set with the peripheral element representation of external elements not included in the set;
An element addition procedure for determining whether or not the external element should be added to the set according to the comparison result of the comparison procedure between the element sets, and adding if it should be added;
27. The image processing program according to any one of appendices 24 to 26, wherein:
(Supplementary Note 28) In the set expression generation procedure, the computer is highly likely to include all or all of the constituent elements constituting the image in the set range. 28. The image processing program according to appendix 27, wherein the image processing program is expressed by a continuous value or a multivalued value including a ternary value.
(Supplementary Note 29) The inter-element set comparison procedure is performed on the computer.
The features in the elements included in all of the constituent elements constituting the image or in a predetermined range expressed by the set expression generation procedure and the peripheral element characterization procedure are integrated in each element,
By summing up the integration results for all applicable elements,
29. The image processing program according to appendix 27 or 28, wherein the element is compared with the set.
(Supplementary Note 30) The image processing program is stored in the computer.
An image segmentation procedure for segmenting the image in advance;
A subdivided image input procedure that gives each of the subdivided images as input images of the initial nucleus selection procedure, the peripheral element representation procedure, and the set generation procedure;
An integrated image input procedure for giving an image having an area configured in the subdivided image as a component as an input of each of the initial nucleus selection procedure, the peripheral element representation procedure, and the set generation procedure;
30. The image processing program according to any one of appendices 21 to 29, wherein the program is further executed.

According to the present invention, it is possible to perform image processing that makes it possible to extract silhouette information of an object shape from an image and specify the position and shape of the object shape. It is applicable to such uses.

DESCRIPTION OF SYMBOLS 10 Control part 20 Image information acquisition part 30 Image information storage part 40 Aggregate information storage part 50 Processing result output part 101 Initial nucleus selection part 101-1 Component element characterization part 101-2 Element feature comparison part 101-3 Optimal element selection part DESCRIPTION OF SYMBOLS 102 Peripheral element expression part 102-1 Peripheral element area | region specific part 102-2 Peripheral element characterization part 103 Set generation part 103-1 Set expression generation part 103-2 Comparison element between element sets 103-3 Element addition part 201 Input image control Section 201-1 Image segmentation section 201-2 Segmented image input section 201-3 Integrated image input section This application is based on Japanese Patent Application No. 2012-206877 filed on September 20, 2012. Claim priority and incorporate all its disclosure here.

Claims

An image processing system for dividing an object or a region of a shape included in an image from the image,
An initial nucleus selecting means for selecting one or more elements, which are starting points for starting calculation necessary for specifying the region, among the components constituting the image, as an initial nucleus;
A peripheral element expression means for characterizing the peripheral element as a peripheral element expression by comparing each of the constituent elements with the peripheral elements around them;
Whether each of the peripheral elements belongs to the same region as the initial nucleus by comparing the characteristics of the constituent elements of the initial nucleus or the region constituting the region and the peripheral element representation of the peripheral element. Set generation means for generating the set of the constituent elements forming the region and dividing the region from the image;
An image processing system comprising:
The initial nucleus selection means includes
Component characterization means for expressing each of the components constituting the image as element features;
Element feature comparison means for comparing the element features;
Optimal element selection means for selecting an optimal element as the initial nucleus from the comparison result of the element characteristics;
The image processing system according to claim 1, further comprising:
The component characterization means relates to the corresponding element and all of the components constituting the image.
i) the maximum value of the difference in feature quantity from the element included in the determined range;
ii) the number of elements whose difference is within a predetermined numerical value; and iii) the difference between the predetermined characteristic and the corresponding element characteristic;
The image processing system according to claim 2, wherein the characterization is performed by any one of the following.
The peripheral element expression means includes:
Peripheral element region specifying means for specifying another constituent element to be a peripheral element for each of the constituent elements constituting the image;
A peripheral element characterizing means for characterizing each peripheral element by comparing each of the constituent elements with the peripheral element;
The image processing system according to claim 1, further comprising:
The peripheral element characterization unit ternarizes the difference value of the feature amount obtained by comparing the corresponding element with the peripheral element determined by the peripheral element region specifying unit, or the difference value by using two types of threshold values. 5. The image processing system according to claim 4, wherein the characterization is performed according to the obtained value.
The threshold for performing ternarization in the peripheral element characterization means is:
6. The method according to claim 5, wherein the determination is performed by performing a normalization process using feature amounts belonging to the image, including distribution information of difference values of feature amounts of the constituent elements constituting the image. The image processing system described.
The set generation means includes:
A set expression generation means for generating a set expression by using a component of the set including the initial nucleus, and a feature of the set generated using the initial nucleus as a starting point;
A comparison means between element sets for comparing the set representation of the set with the peripheral element representation of external elements not included in the set;
An element addition unit that determines whether or not the external element should be added to the set according to a comparison result of the inter-element set comparison unit;
The image processing system according to claim 4, further comprising:
The image processing system includes:
Image segmentation means for segmenting the image in advance;
Subdivided image input means for giving each of the subdivided images as input images of the initial nucleus selection means, the peripheral element representation means, and the set generation means,
An integrated image input means for giving an image having a region formed in the subdivided image as a component, as an input of each of the initial nucleus selection means, the peripheral element representation means, and the set generation means;
The image processing system according to claim 1, further comprising:
An image processing method for dividing a region of an object or a shape included in an image from the image,
An initial nucleus selection step of selecting one or more elements, which are starting points for starting calculation necessary for specifying the region, among the components constituting the image as an initial nucleus;
A peripheral element expression step for characterizing the peripheral element as a peripheral element expression by comparing each of the constituent elements with the peripheral elements around them,
Whether each of the peripheral elements belongs to the same region as the initial nucleus by comparing the characteristics of the constituent elements of the initial nucleus or the region constituting the region and the peripheral element representation of the peripheral element. A set generation step of generating a set of the constituent elements forming the region and dividing the region from the image;
An image processing method comprising:
An image processing program for causing a computer to divide an object or shape area included in an image from the image, wherein the computer
An initial kernel selection procedure for selecting, as an initial kernel, one or more elements that are starting points for starting calculation necessary to identify the region from among the components constituting the image;
A peripheral element expression procedure for characterizing the peripheral element as a peripheral element expression by comparing each of the constituent elements with their peripheral peripheral elements;
Whether each of the peripheral elements belongs to the same region as the initial nucleus by comparing the characteristics of the constituent elements of the initial nucleus or the region constituting the region and the peripheral element representation of the peripheral element. A set generation procedure for generating a set of the constituent elements forming the region and dividing the region from the image;
An image processing program for executing