KR20190122966A - Method for producing unique ID of object - Google Patents

Abstract

The present invention relates to a method for generating a unique ID of an object. More specifically, the method for generating a unique ID of an object can generate a unique ID capable of identifying all objects such as not only people but also animals, things, or the like by generating a unique ID considering characteristics of the object extracted from a plurality of object images, does not compare an input object image with a registration object image by generating a unique ID consisting of numbers generated from the object image. Accordingly, in the method for generating a unique ID of an object, the object image of all objects is not necessarily registered and stored separately, and all objects can be identified by comparing the unique ID generated from the input object image with the registered unique ID.

Description

Method for producing unique ID of object}

The present invention relates to a method of generating a unique ID of an object. More specifically, by generating a unique ID in consideration of characteristics of an object extracted from a plurality of object images, a unique ID capable of identifying not only a person but also an animal, an object, and the like can be identified. It does not compare the input object image and the registered object image by generating a unique ID consisting of numbers generated from the object image, so there is no need to register and store the object image of all objects separately, and generates a unique ID generated from the input object image. The present invention relates to a method of generating a unique ID of an object capable of identifying all objects by comparing with a registered unique ID.

In authentication or security, IDs are used to identify users or objects. For example, IDs are used in a variety of real-world and industrial applications, such as building access, banking requiring security, authentication of the use of smart devices, and identification of objects.

Such an ID is usually composed of a plurality of characters and symbols arbitrarily set by a user or an administrator, or it is common to use the user's biometric information itself as a unique ID.

In the case of an ID consisting of letters and symbols, the user has a problem that the user must remember and use the ID, and the ID is easily exposed to a third party and exploited.

On the other hand, when the biometric information itself is used as a unique ID, the biometric information of all users is obtained by acquiring the biometric information and comparing the obtained biometric information with previously stored biometric information itself to authenticate the user whether the biometric information matches. The information must be stored and registered, and the user's biometric information is extremely personal and has a problem that can cause enormous damage when exposed to a third party.

In addition, the conventional ID is for identifying the user, which is the user generating the ID with symbols and characters for authentication or registration, and does not reflect the characteristics of the user. Furthermore, the conventional ID has a problem that it cannot be generated as an ID for identifying all objects including the animal or the thing by reflecting the characteristics of the animal or the thing.

The present invention is to solve the problems of the above-described conventional ID generation method, an object of the present invention is to provide a method for generating a unique ID of the object reflecting the characteristics of the object.

Another object of the present invention is to provide a method of generating a unique ID that can identify all objects such as animals, objects as well as humans by generating a unique ID in consideration of characteristics of an object extracted from an object image.

Another object of the present invention is to extract a feature point representing an object from a plurality of object images to remove a background area and to generate an ID of a unique object that is invariant from unique feature points common to a plurality of object images among the extracted feature points. To provide a way.

Another object of the present invention is to provide a method for accurately generating a unique ID of an object with a small amount of calculation from a descriptor vector for a feature point of an object image generated by using a Scale Invariant Feature Transform (SIFT) algorithm.

In order to achieve the object of the present invention, the method for generating a unique ID according to the present invention estimates the lighting components in a plurality of different image of the object photographing the same object and normalizes the lighting for each of the plurality of object images based on the estimated lighting component Generating a decoded image, extracting a keypoint for each illumination normalized object image using a Scale Invariant Feature Transform (SIFT) algorithm, and generating a descriptor of the extracted feature point; and a feature point of each illumination normalized object image. Extracting unique feature points of objects matching each other from the illumination normalized object image for each object image from the distance between the feature points based on the descriptor, and extracting the unique feature points of the objects matched with each other. Candidate of each illumination normalized object image generated from the vector matrix of the descriptor And generating a unique ID of the object from the ID.

Preferably, the illumination normalized object image according to an embodiment of the present invention inverts the object image to generate an inverted object image, and inverse estimation by applying an average filter to the inverted object image to extract low frequency illumination components Calculating an illumination component; reversing the inverse estimated illumination component to calculate an estimated illumination component; generating an illumination normalized inversion object image by combining the estimated illumination component and the object image; Generating an inverted smoothing image in which the pixel value of the illumination normalized inverted object image is diffusely distributed by histogram equalization of the object image, and generating an illumination normalized object image by reversing the inverted smoothing image. Characterized in that.

Preferably, the step of extracting the unique feature point of the object according to an embodiment of the present invention is based on the feature point descriptor for the feature point group consisting of the feature points of the two illumination normalized object images selected from the plurality of illumination normalized object images. Extracting candidate unique feature points that match each other in the feature point group from the distance between the feature points and the unique feature points that are commonly matched in all the feature point groups based on candidate unique feature points extracted from all the feature point groups generated from the plurality of illumination normalized object images. It characterized in that it comprises a step of extracting.

In the step of extracting the candidate unique feature points, the candidate unique feature points are extracted by Equation (1) below.

[Equation 1]

Where v _1n is a feature point of the first illumination normalized object image, v _2i , v _2j are two feature points of the position closest to v _1n among the feature points of the second illumination normalized object image, and d _R is an extraction weight , V _1n , when the equation (1) is satisfied. v _2i is characterized as being extracted as a candidate unique feature point.

Preferably, the step of generating a unique ID according to an embodiment of the present invention comprises the largest number and the largest in the eight-dimensional principal component of each partial vector matrix generated from the vector matrix of descriptors for the unique feature points of each illumination normalized object image. Determining a small number of positions, and generating a candidate ID for each illumination normalized object image, the first and second values representing the largest and smallest positions of each partial vector matrix; And generating unique IDs of objects having a common number from candidate IDs for each lighting normalized object image.

Preferably, extracting the feature point according to an embodiment of the present invention comprises generating a scale space image blurred by a Gaussian function having a different level of variance, and from a difference between adjacent images in the scale space image. Generating a Gaussian difference image; and determining a maximum or minimum pixel value of a neighboring pixel adjacent to the current pixel of the Gaussian difference image, and a next or next pixel value adjacent to the current pixel among the upper and lower adjacent neighboring Gownian difference images. The method may include determining a candidate feature point, and extracting the feature point by removing the unstable candidate feature point from the candidate feature point.

Calculating the azimuth and size between the feature point and the feature point adjacent pixel using the adjacent pixels around the feature point, determining a reference orientation of the feature point from the calculated histogram of the azimuth angle, and surrounding the feature point based on the reference orientation. And generating a feature point descriptor represented by a gradient direction and a size of a local area located at.

In this case, the noise normalized object images of the three object images are obtained by removing noise from three different object images photographing the same object.

The method of generating a unique ID of an object according to the present invention has various effects as follows.

First, the method of generating a unique ID of an object according to the present invention generates a unique ID in consideration of the characteristics of an object extracted from an object image, thereby generating a unique ID capable of identifying all objects such as not only a person but also an animal.

Second, the method of generating a unique ID of an object according to the present invention extracts a feature point representing an object from a plurality of different object images and again generates a unique ID of the object from unique feature points common to the plurality of object images among the feature points. Unique ID of an object can be generated accurately by reflecting its characteristics.

Third, the method of generating a unique ID of an object according to the present invention does not compare the input object image and the registered object image by generating a unique ID composed of numbers generated from the object image, and thus needs to separately store and store the object image of all objects. No object can be identified by comparing the unique ID generated from the input object image with the registered unique ID.

Fourth, the method for generating a unique ID of an object according to the present invention generates a unique ID by using a descriptor vector for a feature point of an object image generated by applying a scale invariant feature transform (SIFT) algorithm, thereby accurately identifying the object with a small amount of computation. You can create an ID.

1 is a functional block diagram illustrating an apparatus for generating a unique ID according to an embodiment of the present invention.
2 is a view for explaining an example of a normalized image generating unit according to the present invention.
3 is a functional block diagram illustrating an example of a unique feature point generation unit according to the present invention.
4 is a flowchart illustrating a method for generating a unique ID of an object according to an embodiment of the present invention.
5 is a diagram for describing an example of generating a ginsian difference image.
FIG. 6 is a diagram for describing an example of determining a feature point having a maximum or minimum pixel value in a next adjacent pixel adjacent to a current pixel in a current image.
FIG. 7A illustrates a gradient direction and a size of a local area located around a SIFT feature point, and FIG. 7B illustrates a gradient direction and a size after applying a Gaussian weight.
8 is a flowchart for explaining an example of the step of extracting a unique feature point according to the present invention.
9 illustrates an example of unique feature points extracted from the first to third illumination normalization object images.
10 is a flowchart illustrating an example of generating a unique ID according to the present invention.
FIG. 11 illustrates an example of a feature descriptor descriptor vector matrix for N unique feature points in an illumination normalized object image.
FIG. 12 illustrates an example of a first value and a second value for each illumination normalized object image.

Technical terms used in the present invention are merely used to describe specific embodiments, it should be noted that it is not intended to limit the present invention. In addition, the technical terms used in the present invention should be interpreted as meanings generally understood by those skilled in the art unless the present invention has a special meaning defined in the present invention, and is excessively comprehensive. It should not be interpreted in the sense of or in the sense of being excessively reduced. In addition, when a technical term used in the present invention is an incorrect technical term that does not accurately express the spirit of the present invention, it should be replaced with a technical term that can be properly understood by those skilled in the art.

Also, the singular forms used in the present invention include plural forms unless the context clearly indicates otherwise. In the present invention, terms such as “consisting of” or “comprising” should not be construed as necessarily including all of the various components or steps described in the invention, and some of the components or some of the steps are included. It should be construed that it may not be, or may further include additional components or steps.

In addition, it should be noted that the accompanying drawings are only for easily understanding the spirit of the present invention and should not be construed as limiting the spirit of the present invention by the accompanying drawings.

Hereinafter, a method of generating a unique ID according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a functional block diagram illustrating an apparatus for generating a unique ID according to an embodiment of the present invention.

Referring to FIG. 1, the image capturing unit 110 captures an image of an object. Here, the object photographed through the image capturing unit 110 is a target having different characteristics for each object, for example, a human face, an iris, a vein pattern, etc. may be an object, and in the field to which the present invention is applied. Accordingly, the face, vein pattern, iris, etc. of a pet such as a dog or a cat may be an object, or an object having different characteristics from each other may be an object. The image capturing unit 110 acquires a plurality of object images by photographing the objects with a time difference from each other. Preferably, the image capturing unit 110 acquires three object images by photographing the object three times with a time difference from each other.

As an example of the image capturing unit 110, a near-infrared ray irradiated from an illumination unit configured to arrange a plurality of near-infrared LED lamps in a plurality of columns and rows passes through or reflects an object, and captures an object image by using light provided. A charge coupled device or complementary metal-oxide semiconductor (CMOS) camera may be used.

The normalized image generating unit 130 estimates an illumination component in each of the plurality of object images obtained from the image capturing unit 110 and generates an illumination normalized image for each of the plurality of object images based on the estimated illumination component. The image capturing unit 110 acquires an object image by using an infrared light source. The infrared light source may be concentrated and irradiated to a specific part of an object according to an arrangement position of the infrared light source. The lighting component in is unbalanced.

The feature point generator 150 extracts a SIFT keypoint representing an object feature from each illumination normalized image by using a scale invariant feature transform (SIFT) algorithm and generates a descriptor of the extracted SIFT feature point. The descriptor of the extracted feature point is the gradient direction and size of the local area located around the SIFT feature point and the position information of the SIFT feature point.

The unique feature point generation unit 170 generates unique feature points that are common to each other in each illumination normalized image by determining feature points that match each other in each illumination normalized image from the distance between the feature points based on the feature point descriptor of each illumination normalized image. . Here, the unique feature points are common feature points that are matched with each other in each lighting normalized image. By removing feature points that do not coincide with each other among feature points of each lighting normalized image, only an object region excluding a background region may be extracted from the lighting normalized image. .

The unique ID generating unit 190 generates a candidate ID from a vector matrix of descriptors for unique feature points for each lighting normalized image, and generates a unique ID of an object from candidate IDs common to each other in candidate IDs of each lighting normalized image. .

2 is a view for explaining an example of a normalized image generating unit according to the present invention.

Referring to FIG. 2, the first inverting unit 131 inverts the object image I (x, y) to generate an inverted image I '(x, y), and estimates an illumination component. 133 calculates an inverse estimated illumination component R (x, y) by convolving the average filter to extract the low frequency illumination component. Since the lighting component appears as a low frequency component in the object image, the average filter is convoluted to the inverted original image to blur the inverted original image, thereby estimating the lighting component in the inverted original image. Here, the average filter has a size of m × n and is convolved with an M × N size inverted original image to generate an inverse estimated illumination component.

The second inverting unit 135 inverts the inverse estimated illumination component R (x, y) to calculate the estimated illumination component R '(x, y).

The illumination normalizer 137 sums the estimated illumination components R '(x, y) and the inverted image I' (x, y) to maintain pixel values between 0 and 255, and Set to 0 and set the pixel value of 255 or more to 255 to generate the illumination normalized inverted image I '' (x, y), and the smoothing unit 158 generates the illumination normalized inverted image I '' (x, y)) is histogram equalized to diffusely distribute the pixel values of the illumination normalized inverted image I '' (x, y). Here, the histogram of the illumination normalized inverted image having poor contrast distribution is uniformly distributed through histogram smoothing, and an illumination normalized object image maximizing contrast is generated.

The third inverting unit 139 inverts the inverted smoothing image to generate the illumination normalized object image E (x, y).

When a pixel value of an image has a value of 0 to 255, the conventional lighting normalization method subtracts an estimated illumination component generated by applying an average filter to an input image from an input signal, and adds 255 to the subtracted value and adds the sum. The illumination was normalized by dividing the value by 1/2. The normalized image generator according to the present invention sums the estimated illumination components R '(x, y) and the inverted image I' (x, y) to maintain pixel values of 0 to 255 and to maintain pixel values of 0 or less. Is set to 0 and a pixel value of 255 or more is set to 255, so that the object image can be filtered more clearly and the accuracy of SIFT feature point extraction and matching can be improved.

3 is a functional block diagram illustrating an example of a unique feature point generation unit according to the present invention.

The group generator 171 generates a group from two different illumination normalized object images selected from the plurality of illumination normalized object images. That is, when generating three illumination normalized object images E1, E2, and E3 from three object images, the first group E1, E2 is generated from two illumination normalized object images selected from three illumination normalized object images. ), Second groups E1 and E3, and third groups E2 and E3.

The candidate unique feature point generator 173 generates candidate unique feature points that match each other in each group based on the distance between the feature points of the illumination normalized object image constituting each group, and the unique feature point generator 175 generates each group. Generates unique feature points that are commonly matched with each other in all groups among candidate unique feature points that are matched with each other.

4 is a flowchart illustrating a method for generating a unique ID of an object according to an embodiment of the present invention.

Referring to FIG. 4, in detail, an illumination component is estimated from the obtained plurality of object images of the same object, and an illumination normalized image is generated for each of the plurality of object images based on the estimated illumination component (S10).

A feature point of an object is extracted from each illumination normalized image using a SIFT algorithm, and a description of a feature point of each generated illumination normalized image is generated (S50).

Looking at the step of extracting the feature point in more detail, the convolution of the Gaussian function and the peaceful illumination normalized finger vein image having the feature points with different levels of variance and blur the scaled spatial image to the following equation (1) Create like this

[Equation 1]

L (x, y, σ) = G (x, y, σ) * E (x, y)

Where G (x, y, σ) is a Giancian function, E (x, y) is a smoothed illumination normalized object image, and L (x, y, σ) is a Gaussian function with different levels of variance (σ). It means a blurred image.

In order to find a stable feature point in the scale space, a Gaussian difference image (DOG) is generated from the difference of adjacent images in the scale space image as shown in Equation (2) below. 5 is a diagram for describing an example of generating a ginsian difference image.

[Equation 2]

DOG = (G (x, y, kσ) * E (x, y) -G (x, y, σ) * E (x, y))

= L (x, y, kσ) -L (x, y, σ)

A feature point having a maximum or minimum pixel value is determined in a neighboring pixel adjacent to a current pixel of a Gaussian difference image and a next neighboring pixel adjacent to a current pixel among upper and lower neighboring neighboring neighboring difference images.

Referring to FIG. 6, a gaussian difference image is used to compare eight pixels in the current image and nine pixels in an adjacent image. Assuming that the 'X' position is the current reference pixel, a total including pixels included in a predetermined region (for example, 3x3) at the same position in the Gaussian difference image adjacent to the neighboring pixel with respect to the 'X' By comparing pixel values of 26 pixels, if the reference pixel is the maximum or minimum value, the reference pixel is registered as a feature point. Feature points are found by repeating the above process using Gaussian difference images for each octave of scale space.

Here, the function m (x, y) for the magnitude value of the SIFT feature point (X (x, y)) and the function θ (x, y) for the direction value are expressed in Equations (3) and (4) below. Is calculated by

[Equation 3]

[Equation 4]

Even though there is rotation of the image, the pixel value around the SIFT feature point (X ') does not change, so the SIFT feature point whose size (m (x, y)) and direction (θ (x, y)) are determined based on the pixel value Since the direction of the SIFT feature is rotated as the image rotates, the reference direction of the SIFT feature is an important criterion when generating a feature descriptor that does not change even with the rotation of the image.

Meanwhile, referring to the method of generating a feature point descriptor in detail, a keypoint descriptor is generated based on the feature point reference direction.

Compute the gradient direction and size of the local area around the SIFT feature point. The gradient direction and size of the local region may be calculated by using Equations (3) and (4).

In order to obtain an invariant descriptor for image rotation, the gradient direction and the coordinates of the descriptor are rotated based on the SIFT feature point direction, and the rotated gradient is given a Gaussian weight around the SIFT feature point. Gaussian weights are applied to prevent small changes in the area around the SIFT feature points from sensitively changing the direction and size of the gradient. Gaussian weights are assigned around SIFT feature points to emphasize gradients around the feature points and minimize errors.

FIG. 7A illustrates a gradient direction and a size of a local area located around a SIFT feature point, and FIG. 7B illustrates a gradient direction and a size after applying a Gaussian weight.

Gaussian weighted gradients are reconstructed into 4x4 arrays, with each array having eight directions. This directional histogram consists of 128 (8x4x4) vectors. The normalization process can be performed to make a 128-dimensional vector a light-resistant descriptor. The 128-dimensional vector generated through this process becomes a feature descriptor that represents the SIFT feature.

Referring to FIG. 4 again, a method of generating a unique ID of an object according to the present invention is based on a feature point descriptor of each light normalized object image from a distance between feature points from a background region of the light normalized object image for each object image. The unique feature points of the objects matching each other are removed by removing the feature points (S70).

Generate a candidate unique ID for each lighting normalized object image from the 128 vector matrix of feature descriptors for the unique feature points of each lighting normalized object image, and use the common IDs for the candidate unique IDs for each lighting normalized object image. Generate a unique ID for the object (S90).

8 is a flowchart for explaining an example of the step of extracting a unique feature point according to the present invention.

Referring to FIG. 8, the feature point group is generated as the feature points of the two illumination normalized images selected from the plurality of illumination normalized images (S71). That is, when three lighting normalized object images E1, E2, and E3 are generated from three object images, the first group E1 is generated from two different lighting normalized object images selected from three lighting normalized object images. , E2), second groups E1 and E3, and third groups E2 and E3 are generated.

Based on the distance between the feature points of the illumination normalized object image constituting each feature point group, candidate unique feature points matching each other in each group are extracted (S73). For example, the feature point of the first illumination normalized object image based on the distance between the feature point of the first illumination normalized object image E1 and the feature point of the second illumination normalized object image E2 constituting the first group; A candidate unique feature point common to each other between the second illumination normalized object images is extracted.

Preferably, candidate intrinsic feature points matching each other in each group are calculated as in Equation 5 below.

[Equation 5]

Where v _1n is a feature point of the first illumination normalized object image, v _2i , v _2j are two feature points of the position closest to v _1n among the feature points of the second illumination normalized object image, and d _R is an extraction weight ,

If the equation (5) is satisfied v _1n , v _2i is extracted as candidate unique feature points.

For all of the generated feature point groups, candidate unique feature points of each feature point group are extracted in the same manner as described above.

Unique feature points that are matched with each other in common from all the feature point groups among candidate unique feature points that match each other in each feature point group are extracted (S75). That is, as shown in FIG. 9, when three illumination normalized images are obtained from three object images, the feature points of the first illumination normalized image E1 and the feature points of the third illumination normalized image E2 And common feature points that are common between the feature points of the third illumination normalized image E3.

10 is a flowchart illustrating an example of generating a unique ID according to the present invention.

Referring to FIG. 10, the position of the largest and smallest numbers in the 8-dimensional principal component of the partial vector matrix generated from the vector matrix of the feature point descriptors for the unique feature points of each illumination normalized object image is determined. S91). As shown in Figure 11, when the first jomyeong N unique feature points in the normalized object image _{(V 1 -1, V 1-2,} ...., V 1 -N) are extracted, each unique feature point In the above-described method, a 128-dimensional feature descriptor vector matrix is generated. The 128-dimensional feature descriptor vector matrix is divided into 8 to generate a partial vector matrix P _i and calculates an average value of the magnitudes of the principal components for each partial vector matrix. In this case, the position of the largest number and the smallest number is determined from each average value.

)과의 행렬곱(

)으로 얻어진 행렬의 고유값과 고유벡터를 계산하여 그 고유벡터들로부터 부분 벡터 행렬(P_i)의 열벡터들의 주성분을 계산하는 방법이다. A partial vector matrix (P _i ) is generated by dividing a 128-dimensional feature point descriptor vector matrix into eight, and the largest and smallest positions of each principal component vector are determined by calculating an eight-dimensional principal component vector from each partial vector matrix. do. Here it means the main component is the matrix-vector portion (P _i) average vector or vector component analysis made by the absolute value of the eigenvector corresponding to the largest number of the eigenvalues obtained through the (PCA, Principle Component Analysis) of the. Principal component analysis is based on the matrix P _i ,

Matrix multiplication with

) To calculate the eigenvalues and eigenvectors of the resulting matrix is a method of calculating the principal component of the column vector of the matrix-vector portion (P _i) from its own vector.

In FIG. 11, the mean value of the principal component size of the largest size of the partial vector matrix P ₁ is 4.0 and the mean value of the smallest principal component size is 2.6. The mean value 4.0 is located eighth and the mean value 2.6 is located third. The mean values of the principal component sizes of the partial vector matrices generated from the feature descriptor descriptor vector matrices of the unique feature points that are common to each other in the plurality of illumination normalized images may be different from each other. It can be seen that the order is almost the same.

Referring back to FIG. 10, a candidate ID for each illumination normalized object image, which is composed of first and second values representing the largest and smallest positions of each partial vector matrix, is generated (S93). . As described above with reference to FIG. 11, in the first partial vector matrix P ₁ , the first and second values are 8 and 3, respectively, and the first and second values are determined for the remaining partial vector matrix. A candidate ID including a first value and a second value for the vector matrix is generated. 12 illustrates an example of a first value and a second value for each illumination normalized object image. A candidate ID 1 including first and second values of all partial vector matrices of the first illumination normalized image is illustrated in FIG. (ID ₁ ), a candidate ID (ID ₂ ) consisting of first and second values of all partial vector matrices of the second illumination normalized image, and first values and first values of all partial vector matrices of the third illumination normalized image A candidate ID consisting of two values (ID ₃ ) is shown below.

ID ₁ = (83, 71, 57, 34, 36, 62, 35, 41, 34, 12, 47, 65, 47, 21, 65, 27)

ID ₂ = (83, 71, 57, 35, 36, 52, 35, 41, 34, 12, 47, 65, 47, 23, 65, 27)

ID ₃ = (83, 71, 57, 35, 36, 52, 35, 41, 34, 12, 47, 65, 47, 24, 65, 27)

In operation 95, a unique ID of an object having a number common to each other is generated from candidate IDs for each lighting normalized object image. That is, when the candidate ID 1 (ID ₁ ) of the first illumination normalized image, the candidate ID (ID ₂ ) of the second illumination normalized image, and the candidate ID (ID ₃ ) of the third illumination normalized image match each other, If the numbers remain the same and do not match with each other, it is set to 0 so that candidate ID 1 (ID ₁ ) of the first illumination normalized image, candidate ID (ID ₂ ) of the second illumination normalized image, and third illumination normalization are set. A unique ID of the object is generated from the candidate ID (ID ₃ ) of the captured image.

According to an embodiment, when one of the candidate IDs generated from each of the plurality of illumination normalized object images does not match each other, it may be set to zero. In another embodiment, when all candidate IDs generated from the plurality of illumination normalized object images are different from each other, the candidate IDs may be set to 0. If there is a matching number, the unique IDs may be set to the matching numbers.

In FIG. 12, a candidate ID 1 (ID ₁ ) of the first illumination normalized image, a candidate ID (ID ₂ ) of the second illumination normalized image, and a candidate ID (ID ₃ ) of the third illumination normalized image are generated. Unique ID (I) according to the example is as follows.

 ID = (83, 71, 57, 30, 36, 02, 35, 41, 34, 12, 47, 65, 47, 20, 65, 27)

Meanwhile, in FIG. 12, candidate ID 1 (ID ₁ ) of the first illumination normalized image, candidate ID (ID ₂ ) of the second illumination normalized image, and candidate ID (ID ₃ ) of the third illumination normalized image are generated. Unique ID (I) according to another embodiment is as follows.

 ID = (83, 71, 57, 35, 36, 52, 35, 41, 34, 12, 47, 65, 47, 20, 65, 27)

Meanwhile, the above-described embodiments of the present invention can be written as a program that can be executed in a computer, and can be implemented in a general-purpose digital computer that operates the program using a computer-readable recording medium.

The computer-readable recording medium may be a magnetic storage medium (for example, a ROM, a floppy disk, a hard disk, etc.), an optical reading medium (for example, a CD-ROM, DVD, etc.) and a carrier wave (for example, the Internet). Storage medium).

Although the present invention has been described with reference to the embodiments shown in the drawings, this is merely exemplary, and it will be understood by those skilled in the art that various modifications and equivalent other embodiments are possible. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

110: image capturing unit 130: normalized image generating unit
150: feature point generator 170: unique feature point generator
190: unique ID generation unit 131: first inverting unit
133: lighting component extraction unit 135: second inversion unit
137: lighting normalization unit 138: smoothing unit
139: third inverting unit 171: group generating unit
173: candidate unique feature point generator 175: unique feature point generator

Claims (8)

Estimating an illumination component from a plurality of different object images photographing the same object and generating an illumination normalized image for each of the plurality of object images based on the estimated illumination component;
Extracting a keypoint for each of the illumination normalized object images using a scale invariant feature transform (SIFT) algorithm, and generating a descriptor of the extracted feature point;
Extracting unique feature points of objects matching each other by removing feature points of a background area from the illumination normalized object image for each object image from the distance between the feature points based on the feature point descriptor of each illumination normalized object image;
Generating a unique ID of the object from an ID candidate of each lighting normalized object image generated from a vector matrix of descriptors for the unique feature points of each lighting normalized object image. . The method of claim 1, wherein the illumination normalized object image
Inverting the object image to generate an inverted object image;
Calculating an inverse estimated illumination component by applying an average filter to the inverted object image to extract a low frequency illumination component;
Reversing the inverse estimated illumination component to calculate an estimated illumination component;
Generating an illumination normalized inverted object image by adding the estimated illumination component and the object image to each other;
Generating a inverted smoothed image in which histogram equalization of the illumination normalized inverted object image is performed to diffusely distribute pixel values of the illumination normalized inverted object image; And
Generating an illumination normalized object image by reversing the inverted smoothed image. The method of claim 1, wherein the extracting the unique feature points of the object
Extracting candidate intrinsic feature points that match each other in the feature point group from the distance between the feature points based on the feature point descriptor for the feature point group consisting of the feature points of the selected two illumination normalized object images among the plurality of illumination normalized object images; And
And extracting unique feature points that are commonly matched in all feature point groups based on candidate unique feature points extracted from all feature point groups generated from the plurality of illumination normalized object images. The method of claim 3, wherein in the extracting of the candidate unique feature points, the candidate unique feature points are extracted by Equation (1) below.
[Equation 1]

Where v _1n is a feature point of the first illumination normalized object image, v _2i , v _2j are two feature points of the position closest to v _1n among the feature points of the second illumination normalized object image, and d _R is an extraction weight ,
If you satisfy the equation (1) v _1n , v _2i is a method for generating a unique ID of an object, which is extracted as a candidate unique feature point. The method of claim 3, wherein generating a unique ID
Determining the largest and smallest positions in the eight-dimensional principal component of each partial vector matrix generated from the vector matrix of descriptors for the unique feature points of each illumination normalized object image;
Generating a candidate ID for each illumination normalized object image, comprising a first value and a second value representing the largest and smallest positions of each partial vector matrix;
And generating a unique ID of an object having a number common to each other from candidate IDs of the lighting normalized object images. The method of any one of claims 1 to 5, wherein the extracting the feature points
Generating a scale spatial image blurred by a Gaussian function having a different level of variance;
Generating a Gaussian difference image from the difference of adjacent images among the scale spatial images;
Determining a candidate feature point having a maximum or minimum pixel value in a neighboring pixel adjacent to a current pixel of the Gaussian difference image and a next neighboring pixel adjacent to the current pixel among upper and lower adjacent neighboring gownian difference images; ; And
And extracting the feature point by removing the unstable candidate feature point from the candidate feature point. The method according to any one of claims 1 to 5,
Calculating an azimuth and magnitude between the feature point and the feature point adjacent pixel using the pixels adjacent to the feature point around the feature point;
Determining a reference orientation of the feature point from the calculated histogram of the azimuth angle;
And generating a feature point descriptor expressed in a gradient direction and a size of a local area located around the feature point based on the reference orientation. The method according to any one of claims 1 to 5,
A method of generating a unique ID of an object, characterized by obtaining a lighting normalized object image for each of the three object images by removing noise from three different object images photographing the same object.

Images