JP2000137812A - Individual identifier - Google Patents

Abstract

PROBLEM TO BE SOLVED: To decrease the identification errors while reducing the burden required for the initial dictionary registering jobs. SOLUTION: A dictionary storing part 4 stores the dictionaries of the same ID and different photographing conditions. Each of these dictionaries has a formal registration flag, a temporary registration flag and un-registration flag respectively. A collation part 3 collates an input image with a formally registered dictionary having the same ID as the input image to decide the genuineness of the input image. When a genuine input image is decided at the part 3, a dictionary register part 5 checks whether the dictionary having the same ID and different photographing condition from the input image is temporarily registered or not. If the said dictionary is temporarily registered, the part 5 checks whether the degree of difference between the input image and the temporarily registered dictionary is smaller than a prescribed degree. If the said degree of difference is smaller than the prescribed degree, the part 5 decides the temporarily registered dictionary of the part 4 as a formally registered one.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an individual identification apparatus for identifying individuals based on biological characteristics, and more particularly, to an individual identification device.
For example, the present invention relates to an animal individual identification device for registering data for individual identification of animals and performing collation in an organization requiring individual management of animals such as (race) horses and (pastoral) cows.

[0002]

2. Description of the Related Art Individual identification of thoroughbreds at a racetrack, auction field, or a breeding ranch is carried out under the initiative of a pedigree control agency. The current method of identifying individuals in Japan is based on coat color, vitiligo on the limbs, vitiligo on the head, and curl. In the world, there are other methods based on blood type, branding, tattooing, etc. Reference: "Knowledge of Horses", II-
3. How to identify a horse, P. 153-P. 154).

[0003] In the case of individual identification of cattle, a method using identification nameplates such as collars and ear tags and branding and tattooing has been adopted in general ranches.

[0004] As an individual identification method of an animal that has recently attracted attention, there is a technique using an MC (Micro Chip). This is a chip with a built-in microminiature integrated circuit enclosed in a glass tube, which is implanted in the body of an animal with an injector such as a syringe.In the case of identification, the embedded signal is traced by a non-contact detector to trace the output signal. This is used as information for individual identification.

[0005]

However, the above-mentioned prior art has the following problems to be solved.
In the case of discrimination based on coat color, white mark, vitiligo, and curl, there are many horses with few features and horses with the same feature, so that the identification of the horse is often incomplete. In addition, there is a risk that the branding and tattoos may disappear or be tampered with, and improvements in animal welfare such as pain and local suppuration received by animals have been desired. In addition, although individual identification based on blood type is accurate, there is a disadvantage that a considerable processing time is required for determination and the cost is high.

[0006] The identification nameplate has a risk of being damaged, lost or stolen, and in the case of branding or tattooing, it has a risk of disappearing or being tampered with like a horse. have. In particular, even if this falsification is revealed, it is difficult to prove where the person originally belonged.

[0007] The MC method can be used semi-permanently once it is embedded, and also has advantages such as high convenience. However, the operability of implantation in living animals, pain given to animals at the time of implantation, local reactivity such as swelling, tenderness, suppuration, etc., motor dysfunction and clinical abnormalities in animals, movement of MC in vivo There are many issues in terms of performance, operability of detector, change in detection sensitivity, stability, and reliability. And above all, from the aspect of animal welfare, there are animal officials who are reluctant to adopt the MC method,
A convenient identification method that replaces the MC identification method has been desired.

[0008]

The present invention employs the following structure to solve the above-mentioned problems. <Configuration 1> The same ID is provided for each shooting condition, and
A dictionary storage unit that stores a dictionary having information indicating whether it is formal registration, temporary registration, or unregistered; a shooting condition measurement unit that measures shooting conditions of an input image of a shot individual; and an input of a shot individual The image is collated with the dictionary registered formally in the dictionary storage unit, and if there is a dictionary having a degree of difference smaller than the first threshold, the input image is determined to be true with respect to the ID of the dictionary. If the input image is determined to be true as a result of the collation by the collation unit and the dictionary is not registered in the shooting conditions of the input image, the input image is temporarily registered as a dictionary, and the input image is registered. If the dictionary under the shooting condition is provisionally registered, it is determined whether the difference between the input image and the provisionally registered dictionary is smaller than a second threshold set to a value smaller than the first threshold, If so, a dictionary registration unit that makes the temporary registration dictionary an official registration dictionary Individual identification apparatus, comprising the.

<Structure 2> A dictionary storage unit for storing a formal registration dictionary and a tentative registration dictionary with the same ID, a photographed input image of an individual, and a formal registration dictionary in the dictionary storage unit are compared. If there is a dictionary whose dissimilarity is smaller than the first threshold value, the input image is compared with a matching unit that determines that the dictionary image is true with respect to the ID of the dictionary. If it is determined to be true, the input image is compared with the temporary registration dictionary of the same ID stored in the dictionary storage unit, and the difference between them is set to a value smaller than the first threshold. It is determined whether it is smaller than the second threshold, and if so, the difference between the temporary registration dictionary and the formal registration dictionary of the same ID is larger than the second threshold and smaller than the first threshold. A distance calculation unit that determines whether the value is greater than a third threshold set for the value; With a release calculation unit, when the input image, temporarily registered dictionary and degree of difference not smaller than the second threshold value, stored in the dictionary storage unit an input image as a provisional registration dictionary, and,
When the degree of difference between the input image and the temporary registration dictionary is greater than a third threshold, a dictionary registration unit that registers the temporary registration dictionary as a formal registration dictionary.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below in detail with reference to specific examples. The invention relates to (race) horses,
(Maki) In order to identify animals at an organization that requires individual management of animals such as cattle, for example, register data on iris and iris granules and register data for collation processing.
The present invention relates to a verification device. Hereinafter, in each specific example, a case where a target animal is a horse in particular will be described as an example.

FIG. 2 shows a front view of the eyeball of a horse. As shown in FIG. 2, the horse's eye mainly includes a pupil 10, an iris (iris) 11, and an iris granule 12 between upper and lower eyelids 13.

A transparent hemispherically raised cornea is provided in front of the pupil 10 and the iris 11. The major difference between horse eyes and human eyes is that the pupil 10 is elliptical,
It is to have iris granules 12 unique to horses and ruminants.

Light from the outside passes through the pupil 10 and reaches the retina at the back of the pupil. The iris 11 is a muscle surrounding the pupil 10 and has a function of controlling the amount of light incident on the pupil 10 by contracting and expanding. The iris granules 12 are located between the iris 11 and the pupil 10 and have a shape in which hemispherical raised granules are connected. The iris granules 12 have a black color rich in melanin pigment, and are said to have a function of absorbing dazzling daylight even when the iris 11 contracts.

The size and shape of the iris 11 and the iris granules 12 or the fine wrinkles and irregularities on the surface differ depending on the individual of the horse and the right and left eyes, and there are individual differences. for that reason,
The iris 11 and the iris granules 12 are registered as reference data (hereinafter, referred to as a "dictionary") by imaging the horse's eyes with a camera, and the iris 11 and the iris granule data obtained at the time of the examination are compared and compared with the dictionary. It is possible to identify the target individual.

However, when the photographing condition in the dictionary registration and the photographing condition at the time of collation are largely different, the difference between the dictionary and the input image is large, so that the frequency of errors in which a correct horse cannot be recognized increases, and Since it must be repeated, convenience may be reduced. Here, the photographing condition is a photographing environment such as a biological reaction such as mydriasis and miosis of a pupil due to a difference in illuminance, a photographing angle, focus, and illuminance.

One method of avoiding such an error is to register a plurality of dictionaries of various photographing conditions of the same horse and collate using a dictionary close to the input conditions. However, this takes time to photograph fast-moving horses at various angles, but has limited patience for the horses to stand still, or because they control mydriasis / miosis. Since new equipment is required, the cost is high.

In the present invention, on the other hand, by adding a dictionary automatic addition function to the individual identification device, individual identification processing is performed with a minimum dictionary in the initial operation, and various types of data obtained during the collation are obtained. By sequentially and additionally registering only the images of the shooting conditions which are guaranteed to be correct horse images, the identification error can be reduced while the initial dictionary registration work is reduced. Hereinafter, such an individual identification device of the present invention will be described in detail using specific examples.

<< Specific Example 1 >> In Specific Example 1, a dictionary is provided for each photographing condition. If a dictionary with the same photographing condition as the input image is not registered, the input image is temporarily registered as a dictionary, and the same is repeated once again. When the input image of the condition arrives, it is checked against the input image to confirm that it really matches. If this is confirmed, the temporary registration dictionary is set as the official registration dictionary.

<Configuration> FIG. 1 is a configuration diagram showing a specific example 1 of the individual identification device of the present invention. The specific example 1 includes an imaging unit 1, an imaging condition measurement unit 2, a collation unit 3, a dictionary storage unit 4, a dictionary registration unit 5, and a display unit 6.

The imaging unit 1 comprises an image input unit such as a CCD video camera and an illumination device for certifying an individual.
It captures the image of the horse's eyes as a signal. The image input unit is a part that captures an image of a horse's eye with a camera, and performs a process of converting analog optical information of the image of the horse's eye into a digital electric signal by a sensor. Also, the lighting device is provided to illuminate the horse's eyes in order to capture it with good contrast,
As a light source, light having a wavelength that is hardly perceived by an animal, such as red light, near-infrared light, or infrared light, is used to prevent the horse from dazzling and moving excessively.

The photographing condition measuring section 2 has a function of measuring photographing conditions of an image picked up by the image pickup section 1.
Here, the photographing conditions include conditions such as mydriatic degree, photographing angle, photographing size, illuminance, and focus.

The collation unit 3 collates the image photographed by the imaging unit 1 with a dictionary registered formally in the dictionary storage unit 4, and when there is a dictionary having a degree of difference d1 smaller than the first threshold th1. Has a function of determining that the input image is true with respect to the dictionary ID. That is, the matching unit 3 selects dictionary data of the same shooting condition from the dictionary storage unit 4 and reads the dictionary data into a memory (not shown), compares it with the horse data obtained by the imaging unit 1, and calculates the difference. A function (hereinafter, this difference is referred to as a difference) and a true / false determination function for determining whether or not the horse is the same as the dictionary based on whether the difference is within a threshold value.

The dictionary storage unit 4 is an image database of registered horses. The registered horse IDs, photographing conditions, images, official registration,
It consists of a flag that distinguishes between temporary registration and unregistered.

FIG. 3 is an explanatory diagram showing an example of the dictionary storage section 4. As shown in the figure, the dictionary storage unit 4 is a database that stores a dictionary provided for each shooting condition with the same ID, and each dictionary has a flag indicating whether the dictionary is formally registered, provisionally registered, or not registered. . Further, "-" in the image indicates that the image has not been registered, and "O" indicates that the image has been registered.

Further, as described above, the photographing conditions include the degree of mydriasis, the photographing angle, the photographing size, the illuminance, the focus, and the like.
For example, the imaging conditions A, B, and C are specific conditions such as a value in a range where the mydriatic degree is small to large, and a range of a combination of the mydriatic degree and the imaging angle. Further, the flag
The value is 2 when the dictionary is formally registered, 1 when it is temporarily registered, and 0 when it is not registered.

The dictionary registration unit 5 is a function unit for registering an input image as a dictionary, and includes a flag check function, a temporary registration function of the input image in the dictionary, and a formal dictionary registration function of the temporary registration dictionary. That is, the dictionary registration unit 5 determines that the input image is true as a result of the collation by the collation unit 3 and that the dictionary under the shooting condition of the input image has not been registered (flag = 0). , Temporary registration of input image as dictionary (flag = 1)
On the other hand, if the dictionary under the shooting conditions of the input image is provisionally registered, the difference d2 between the input image and the provisionally registered dictionary
Has a function of determining whether the temporary registration dictionary is a formal registration dictionary (flag = 2) if the second registration threshold is smaller than a second threshold th2 set to a value smaller than the first threshold th1. are doing.

The display unit 6 has a function of displaying the truth / false judgment result / dissimilarity received from the output result of the collation unit 3 and a notification function of dictionary registration received from the output result of the dictionary registration unit 5. Consists of an output device.

In the above specific example, the photographing condition measuring unit 2 to the dictionary registering unit 5 are realized by a processor executing a program corresponding to each function. Further, such a program is provided by being recorded on a recording medium.

<Operation> FIG. 4 is a flowchart showing the operation of this embodiment. The operation will be described below in the order of each step.

[Step S1] The operation of step S1 is as follows.
This is performed by the imaging unit 1. The horse's eyes are illuminated by an illuminator using red, near-infrared, or infrared light as a light source, and imaged by an image input device such as a CCD video camera. The CCD sensor is configured by a two-dimensional array of photoelectric conversion elements, and exposes reflected light of an imaging target for a certain period of time.
The voltage level (analog signal) impressed by the photoelectric conversion is converted into a digital signal by quantizing it at a constant gradation. Thus, a two-dimensional array of image frames is obtained.

[Step S2] Step S2 is an operation performed by the photographing condition measuring unit 2. The photographing condition measuring unit 2 measures photographing conditions such as a mydriatic degree, a photographing angle, a photographing size, an illuminance, and a focus, based on the image obtained from the image pickup unit 1. As a specific measurement method of these, in the image pupil, iris granules, because there is reflection of illumination on the cornea in front of the pupil, the mydriatic degree is the ratio of the vertical width to the horizontal width of the pupil,
The shooting angle can be measured using the position of the reflected light with respect to the center of the pupil, the photographed size can be measured using the area of the reflected light, the illuminance can be measured using the average luminance of the iris granules, and the focus can be measured using the luminance difference between the light and its surroundings. This is not the case.

Here, as the pupil detection method, for example, there is the following method. Since the pupil has a lower density than the iris, the boundary between the pupil and the iris has a large change in density. Therefore, the pupil can be detected by detecting a coordinate sequence at the boundary using a difference filter and then fitting an ellipse that best approximates those points.

Further, as a method for detecting iris granules, for example, the following method can be used. At the upper part of the pupil are iris granules, on which the iris is located. The change in density at the boundary of organs is larger than the change inside each organ. Therefore, first, when a density change detection process represented by the Sobel operator is performed, the pixel value at the boundary has a large value. Thus, the contour of the iris granule can be obtained by obtaining such an area of the pixel value.

Further, the following method can be used for detecting the position of the reflection of the illumination. The reflection of the illumination has a very high density (bright). Therefore, a threshold value can be set in advance, and a pixel region having a density equal to or higher than the threshold value can be extracted as an illumination reflection region.

As a method of detecting the pupil and the reflection of the illumination, for example, a document “Sakaguchi, Nakano, Yamamoto,“ Detection of driver's line of sight while driving a car ”, IEICE
A method described in RU Technical Report, PRU95-28 ”or the like can also be used.

[Step S3] Step S3 and steps S4 to S6 are operations performed by the matching unit 3 and the dictionary storage unit 4.

The collation unit 3 sends an ID number and a photographing condition to the dictionary storage unit 4 in order to select and read a dictionary.
The ID number is input by a horse user from an ID input unit (not shown). The dictionary storage unit 4 searches the list for the data corresponding to the ID number and the photographing condition. If the flag is 2, that is, if it is an officially registered dictionary, the dictionary storage unit 4 sends this to the matching unit 3. On the other hand, if the flag is 0 or 1, a dictionary with a matching shooting condition is not officially registered, and a formal registration dictionary (flag: 2) with a closer shooting condition is sent.

For example, it is assumed that the dictionary storage unit 4 has the configuration shown in FIG. 3, and the order of the photographing conditions is A, B, C, D, E, F,. Then, the input ID number is “1”
And the shooting condition of the input image is “D”. In such a case, the matching unit 3 sends the information of the ID number “1” and the imaging condition “D” to the dictionary storage unit 4. As a result, the dictionary storage unit 4 searches for the data of the shooting condition “D”. Since the data of the shooting condition is provisionally registered, the dictionary of the shooting condition “C”, which is a formal registration dictionary of the closer shooting condition, is retrieved. To the collating unit 3.

[Step S4] The collating unit 3 compares this with the image of the horse to be identified obtained from the imaging unit 1, and calculates the degree of difference. As a calculation method of the degree of difference, a known technique such as an average difference between each pixel of the image of the horse to be identified and the dictionary image and a Euclidean distance can be applied to this specific example.

[Step S5] True / false judgment is made based on the degree of difference. The difference d1 is compared with a predetermined first threshold th1. If the difference is smaller than the threshold, the inspection target is determined to be true, and if not, false. If the result of the authenticity determination is “true”, the process proceeds to the dictionary registration process of step S7, and if the result is “false”, the process proceeds to the rejection process of step S6.

[Step S6] The fact that the identification result is rejection is displayed on the display unit 6, and the process proceeds to the re-identification processing after step S1.

[Step S7] The processing in steps S7 and S8 is performed by the dictionary registration section 5. The dictionary registration unit 5 additionally registers the shortage of the dictionary divided for each shooting condition. This dictionary registration must be performed only when it is determined to be true in the authenticity determination processing in step S5, because the eyes of different horses should not be erroneously registered. In particular,
As checked in the dictionary selection in step S3, the dictionary corresponding to the shooting condition is searched in the dictionary storage unit 4, and if the flag is 2, the dictionary is formally registered, and no additional registration is performed. The process proceeds to the accepting process of step S11. On the other hand, if the flag is 0, that is, if the flag has not been registered, the process proceeds to the temporary registration process in step S9. Further, when the flag is 1, that is, when the temporary registration is performed, the process proceeds to the formal registration determination process from step S8.

[Step S8] If a dictionary that matches the shooting conditions of the input image is provisionally registered, the identifiability of the dictionary is checked as described above. That is, the degree of difference between the input image and the temporary registration dictionary is obtained. If the difference is smaller than the second threshold th2 under a condition more severe than the true / false judgment in step S5, the process proceeds to step S10 to formally adopt the dictionary, otherwise, Proceeds to step S9 to also temporarily register this.

[Step S9] The dictionary is temporarily registered. In the temporary registration, a column of the same ID and the same shooting condition is searched in the items of the dictionary storage unit 4, image data is recorded there, and the flag is rewritten to “1”. For example, if the dictionary with the same ID number and the same shooting condition is “0” or the result of the registration determination in step S8 is equal to or larger than the second threshold, the data of the input image is registered as the temporary registration dictionary (the flag is set to “1”). ”).

[Step S10] If the registration is determined to be OK in step S8, the flag of the temporary registration dictionary is rewritten to “2” and registered as a formal registration dictionary.

As described above, dictionary registration is performed in two stages, such as temporarily registering, temporarily checking whether or not collation can be performed with an input image under the same conditions, and then registering the dictionary formally. This is for preventing erroneous registration of possible dictionaries and suppressing occurrence of erroneous identification.

The processing of steps S5 to S10 is shown in FIG.
This will be described with reference to an example. An input image with an ID number “1” and a shooting condition “D”, an ID number “1” and a shooting condition “C”
When it is determined that the result is “true” as a result of checking the dictionary, the flag of the dictionary with the ID number “1” and the imaging condition “D” is checked. Here, since the flag is 1, it is checked whether the difference d2 between the temporary registration dictionary and the input image is smaller than a second threshold th2. That is, it is checked whether the temporarily registered image and the input image are similar. As a result, if the degree of difference is smaller than the second threshold value, it is determined that they match, and the temporary registration dictionary is set as the official registration dictionary (ID number “1”,
The flag of the dictionary for the shooting condition “D” is changed from 1 to 2.) In this way, it is possible to automatically register a dictionary while preventing erroneous registration.

[Step S11] The fact that the identification result is "accept" is displayed on the display unit 6, and the process is terminated. The above is the description of the operation of this specific example.

<Effects> As described above, according to the first embodiment, in the animal individual identification apparatus, a dictionary is provided for each photographing condition. The input image is temporarily registered as a dictionary, and when an input image with the same conditions comes again, it is checked against it to confirm that it really matches.If this is confirmed, the temporary registration dictionary is officially registered. Therefore, the following effects are obtained.

Since the additional registration of the dictionary can be performed safely and accurately during operation of the identification device, erroneous identification can be reduced by using the dictionary corresponding to various photographing conditions. In addition, the effect of reducing the load on the user when initially registering a large number of dictionaries can be expected.

In the above specific example 1, the collating unit 3 sends the ID
The dictionary storage unit 4 sends back only the formal registration dictionary corresponding to the ID number. However, in addition to this, only the shooting condition is sent from the matching unit 3 and the dictionary storage unit 4 sends the shooting condition. A configuration may be adopted in which formal registration dictionaries of all ID numbers under the same imaging conditions (or similar imaging conditions) are transmitted and collated with these dictionaries.

<< Specific Example 2 >> The individual identification device described in the above-described specific example 1 is provided with the photographing condition measuring unit 2 and registers one dictionary for each predetermined photographing condition. In a specific example 2, an individual identification device capable of automatically registering a dictionary even when it is difficult to measure imaging conditions due to the shape of illumination and environmental light will be described.

In this example, the difference in the photographing conditions between the images is defined by the degree of difference being a certain degree. Specifically, for security, it is determined whether the input images are different from each other with the same ID, and additional registration is performed assuming that the photographing conditions are different due to the difference being equal to or more than a required value in the dictionary. Similar to the first embodiment, the additional registration is performed after performing a collation test on the temporarily registered image to confirm that the image effectively functions as a dictionary.

<Structure> FIG. 5 is a diagram showing the structure of an individual identification device according to a second embodiment of the present invention. As shown in FIG.
The apparatus includes an imaging unit 101, a collation unit 102, a dictionary storage unit 1
03, distance calculation unit 104, dictionary registration unit 105, display unit 1
06, which is different from the first example in that the imaging condition measuring unit 2 is not provided and the distance calculating unit 104 is provided. Further, the functions of the dictionary storage unit 103 and the dictionary registration unit 105 are different.

The image pickup unit 101 has the same configuration as the image pickup unit 1 of the first embodiment. That is, it is composed of an image input section such as a CCD video camera and a lighting device, and outputs an image of a horse eye as a digital electric signal.

The collation unit 102 reads the dictionary data from the dictionary storage unit 103 into a memory (not shown),
A function to calculate the degree of difference by comparing with the horse data obtained in
It has an ID determination function for determining a horse from the dictionary with the minimum difference, and a true / false determination function for determining whether or not the horse is the same as the dictionary based on whether the difference is within the threshold.

The dictionary storage unit 103 is an image database of registered horses, and includes registered horse IDs, images, and flags for distinguishing between formal registration and temporary registration. That is, each dictionary in the dictionary storage unit 103 is a list including an ID number, image data, and a flag, and a plurality of dictionaries are registered for each ID. The flag is 2 when the dictionary has been formally registered and 1 when the dictionary has been provisionally registered.

The distance calculation section 104 has a function of determining a candidate for formal registration from the temporary registration dictionary, and a function of comparing the candidate with the formal registration dictionary and determining whether or not formal registration is possible based on the minimum difference.

That is, the distance calculation unit 104
If it is determined that the input image is true, the input image is compared with a temporary registration dictionary of the same ID stored in the dictionary storage unit 103, and the degree of difference between them is The second set to a value smaller than the first threshold th1
Is determined to be smaller than the threshold th2, and if so, the difference between the temporary registration dictionary and the formal registration dictionary having the same ID is larger than the second threshold and smaller than the first threshold. Threshold value th3 (th2 <th3 <t
h1) It has a function to determine whether it is larger than h1).

The dictionary registration unit 105 has a temporary registration function to the dictionary storage unit 103 and a formal dictionary registration function. That is, when the distance calculation unit 104 determines that the difference between the input image and the temporary registration dictionary is not smaller than the second threshold, the dictionary registration unit 105 stores the input image in the dictionary storage unit 103 as a temporary registration dictionary. If the degree of difference between the input image and the temporary registration dictionary is greater than a third threshold, the temporary registration dictionary is registered as a formal registration dictionary.

The display unit 106 has a function of displaying the result of the judgment and the degree of dissimilarity received from the output of the collation unit 102, and a function of notifying the dictionary registration received from the output of the dictionary registration unit 105. Consists of an output device.

It should be noted that in the above specific example 2, the collating unit 102
The dictionary registration unit 105 is realized by a processor executing a program corresponding to each function. Also,
Such a program is provided by being recorded on a recording medium.

<Operation> FIG. 6 is a flowchart showing the operation of this embodiment. The operation will be described below in the order of each step.

[Step S101] The operation of step S101 is performed by the imaging section 101. The horse's eyes are illuminated by an illuminator using red, near-infrared, or infrared light as a light source, and imaged by an image input device such as a CCD video camera. The CCD sensor is configured by a two-dimensionally arranged photoelectric conversion element, and exposes the reflected light of the imaging target for a certain period of time, and converts the voltage level imprinted by the photoelectric conversion into a digital signal by quantizing it at a certain gradation. Thus, a two-dimensional array of image frames is obtained.

[Step S102] Next, step S1
The operations of 02 to S104 are performed by the matching unit 102 and the dictionary storage unit 103. When the matching unit 102 sends a dictionary read request to the dictionary storage unit 103, the dictionary storage unit 103 sends all the officially registered dictionaries with the flag of 2 to the matching unit 3.

[Step S103] The collation unit 102 calculates the degree of difference between all the dictionaries and the input image, and obtains the minimum degree of difference d1. Although the ID of the input image is unknown, the dictionary having the minimum difference as compared with the dictionary of a plurality of IDs can specify the ID as the same ID. As a method of calculating the degree of difference, a known technique such as an average difference between each pixel of the image of the horse to be identified and the dictionary image and a Euclidean distance can be applied to this specific example.

[Step S104] A decision is made as to the authenticity based on the degree of difference. The dissimilarity d1 is compared with a predetermined threshold th1. If the dissimilarity is equal to or less than the threshold, the inspection target is determined to be true, and if not, false. If the result of the true / false determination is “true”, the flow proceeds to the distance calculation processing in step S106 and subsequent steps. If the result is “false”, the flow proceeds to the rejection processing in step S105.

[Step S105] The fact that the discrimination result is reject is displayed on the display section 106, and step S101 is performed.
The process proceeds to subsequent re-identification processing.

[Step S106] Next step S10
Steps 6 to S109 and S111 are performed by the distance calculation unit 104. Distance calculator 1
No. 04 performs two types of distance calculations for determining a formal registration dictionary candidate from the temporary dictionary and determining whether or not formal registration of the candidate dictionary is possible.

The formal registration candidate is performed only for the image determined to be true in the true / false determination processing in step S104 so as not to erroneously register a different horse eye. First, one temporary registration dictionary is selected and read from the dictionary storage unit 103 to determine a formal registration candidate.

[Step S107] Only temporary registration dictionaries having a sufficiently low degree of difference from the input image are set as candidates for formal registration, thereby ensuring the security of automatic dictionary registration. Therefore, first, the degree of difference d2 between the input image and the temporary registration dictionary read in step S106 is obtained and compared with a predetermined threshold th2. move on. Otherwise, the process proceeds to step S111, and similarly, the next provisional dictionary is selected. At this time, the threshold value d2 is set to be much smaller than the threshold value of the authenticity determination processing in step S104, which is a strict determination condition. This is because, out of the temporarily registered dictionaries, only the dictionary that has been confirmed to well match the same ID is set as a candidate for formal registration.

[Step S108] The temporary registration dictionary that has become a candidate for formal registration reads all the formal registration dictionaries with the same ID from the dictionary storage unit 103 in order to check the degree of difference from the formal registration dictionary with the same ID.

[Step S109] The degree of difference d3 between the temporary dictionary as a formal registration candidate and all formal registration dictionaries is compared with a predetermined threshold th3. If all the differences are larger than the threshold value, the conditions for formal registration are satisfied, and the process proceeds to formal registration processing in step S110. Otherwise, it is determined that the photographing conditions of the candidate dictionary have been officially registered, and the process proceeds to the accepting process of step S113. At this time, the registration of the temporary registration dictionary is canceled, and the temporary registration of the input image is not performed.

[Step S110] The flag of the dictionary in the dictionary storage unit 103 is rewritten to 2, and the formal registration is completed.

[Step S111] If all the temporary registrations have been selected, the process proceeds to step S112. If not, the candidate determination process from step S106 is repeated.

[Step S112] A message indicating that the identification result is accept is displayed on the display unit 106, and the process ends.

The operation example of the formal registration will be described more specifically. FIG. 7 is an explanatory diagram of dictionary registration, and is a diagram illustrating a shooting condition space in which two shooting conditions are plotted on the horizontal axis and the vertical axis. Although there are a plurality of photographing conditions, a two-dimensional image is briefly described here for the sake of explanation. In this space, an image is represented by a point because two shooting conditions have specific values. In addition, the difference between the points in the space indicates the degree of difference in this description because the degree of difference tends to increase as the shooting conditions differ.

In FIG. 6, a black circle indicates a formal registration dictionary, a shaded circle indicates a temporary registration dictionary, and a white circle indicates an input image. These images have the same ID. The circles indicated by dotted lines, solid lines, and broken lines are circles formed by the distances of the thresholds th1, th2, and th3. The temporary registration dictionary is used in the true / false determination process in step S104.
Since the images are selected from images smaller than the threshold th1, they are all within the dotted line.

In the true / false discrimination area formed by the dotted line, the entire photographing condition space cannot be covered, so that an input such as a point p0 having the same ID is erroneously rejected.
The dictionary registration of this specific example is a method for efficiently filling such a space in the space. The candidate determination in step S107 and the registration determination in step S109 of the distance calculation unit 104 are performed in the vicinity of the empty area while confirming the collation. This is processing for officially registering a dictionary.

For example, the collation can be confirmed by the input p1, that is, the dictionary that is included in the solid circle and is a candidate for formal registration is q1. Since q1 is outside the dashed circle for all officially registered dictionaries, it is registered as an officially registered dictionary. As a result, a part of the empty area is filled, and it becomes possible to accept the next input such as p0. On the other hand, since the temporary registration dictionary q2 which becomes a candidate by the input p2 is in the circle of the broken line, even if it is formally registered, the duplication of the area is large.
In this way, even when there is no imaging condition measurement unit, it is possible to automatically add a dictionary while avoiding duplication. The above is the description of the operation of the specific example 2.

<Effects> As described above, according to the specific example 2, as a result of the authenticity determination, the input image is determined to be “true”, and the difference between the input image and the temporary registration dictionary having the same ID number is determined. If the degree of difference is small and the degree of difference between the temporary registration dictionary and the official registration dictionary having the same ID number is equal to or more than a predetermined value, the temporary registration dictionary is set as the official registration dictionary, so that the photographing conditions are not measured. In both cases, additional registration of the dictionary can be performed safely and accurately during operation of the identification device, so that erroneous identification can be reduced by using a dictionary corresponding to various imaging conditions. In addition, the effect of reducing the load on the user when initially registering a large number of dictionaries can be expected.

<Usage Form> In each of the above specific examples, the collation unit and the dictionary have been described as using an image. However, even when collation is performed using various features extracted from the image, the same can be applied. . Also, the case of a horse has been described as an individual to be identified, but the present invention is not limited to this, and other animals or humans can be similarly applied.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a specific example 1 of an individual identification device of the present invention.

FIG. 2 is an explanatory diagram of an eyeball image of a horse.

FIG. 3 is an explanatory diagram illustrating an example of a dictionary storage unit in a specific example 1 of the individual identification device of the present invention.

FIG. 4 is a flowchart showing an operation of a specific example 1 of the individual identification apparatus of the present invention.

FIG. 5 is a configuration diagram showing a specific example 2 of the individual identification device of the present invention.

FIG. 6 is a flowchart showing an operation of a specific example 2 of the individual identification apparatus of the present invention.

FIG. 7 is an explanatory diagram of dictionary registration in a specific example 2 of the individual identification apparatus of the present invention.

[Explanation of symbols]

 2 Photographing condition measurement unit 3, 102 Collation unit 4, 103 Dictionary storage unit 5, 105 Dictionary registration unit 104 Distance calculation unit

Claims (2)

[Claims] 1. A dictionary storage unit for storing a dictionary having the same ID, provided for each photographing condition, and having information indicating whether it is formal registration, temporary registration, or unregistered, and an input image of a photographed individual A photographing condition measuring unit for measuring the photographing conditions of the above, a photographed individual input image is collated with a dictionary registered formally in the dictionary storage unit, and a dictionary having a degree of difference smaller than the first threshold value exists. In the case, the input image is determined to be true with respect to the ID of the dictionary, and as a result of matching by the matching unit, the input image is determined to be true, and shooting of the input image is performed. If the dictionary under the condition has not been registered, the input image is temporarily registered as a dictionary.If the dictionary under the shooting condition of the input image is temporarily registered, the input image and the temporarily registered dictionary are registered. Is set to a value smaller than the first threshold value. And second or determined smaller than the threshold value, the If so, the individual identification apparatus is characterized in that a dictionary registration unit for the temporary registration dictionary officially registered dictionary. 2. A dictionary storage unit for storing a formal registration dictionary and a tentative registration dictionary with the same ID, an input image of a photographed individual, and a dictionary registered formal in the dictionary storage unit. If there is a dictionary whose dissimilarity is smaller than the first threshold value, the input image is compared with a matching unit that determines that the ID of the dictionary is true. If it is determined that the image is true, the input image is compared with the temporary registration dictionary of the same ID stored in the dictionary storage unit, and the difference between the input images is smaller than the first threshold. It is determined whether it is smaller than the second threshold set in the above, if it is, the degree of difference between the temporary registration dictionary and the formal registration dictionary of the same ID is larger than the second threshold, and , From a third threshold value set to a value smaller than the first threshold value A distance calculating unit that determines whether the input image is larger than the second image. If the difference between the input image and the temporary registration dictionary is not smaller than a second threshold, the input image is used as the temporary registration dictionary. A dictionary registration unit that stores the temporary registration dictionary as a formal registration dictionary when the degree of difference between the input image and the temporary registration dictionary is greater than the third threshold value. An individual identification device characterized by the above-mentioned.