KR20220020640A - Face Recognition System unaffected disturbing light - Google Patents

Abstract

The present invention relates to a face recognition system which is strong against disturbance light. Provided is a face recognition system, which includes: a camera for photographing a user's face; and at least one polarizing filter determined according to an illuminance environment of the face recognition system and mounted on a front end of the camera. Accordingly, it is possible to prevent a decrease in a face recognition rate due to the disturbance light.

Description

Face Recognition System unaffected disturbing light

The present invention relates to a face recognition system that is robust against disturbance light.

The disturbance light referred to in the face recognition system is a generic term for light incident from the light source to the camera in the forward direction, light incident from the light source reflected by a reflector and entering the camera, and light that is scattered from the same reflector and incident.

As a specific example of disturbance light, it is classified into DC disturbance light such as sunlight or incandescent light, and AC disturbance light such as inverter fluorescent lamp or LED light. In particular, when DC disturbance light is incident on the camera, the photoelectric sensor rises to the saturation voltage level, causing serious noise in the camera image.

Because AC disturbance light has a frequency component similar to the self-luminescence period required for detection by the photoelectric sensor, white burnout occurs in the camera image.

Therefore, in order to develop a face recognition system that is robust to disturbance light, it is essential to understand the characteristics of light and to secure a countermeasure against the conditions of various light sources such as illuminance, luminance, incident angle, and scattered light.

KR 10-887183 B1

The present invention provides a face recognition system robust to disturbance light that can efficiently cope with disturbance light that occurs frequently in the operation of the face recognition system exposed to various environments.

In addition, the present invention minimizes the interference of extraneous light by using polarized light used for LCD and OLED and a wide range of applications, and at the same time, automatically attaches and detaches the polarizing filter by reflecting critical conditions, such as the intervention of the polarizing filter according to various lighting conditions. Alternatively, a face recognition system that can be manually attached or detached is provided.

In order to achieve the above object, the face recognition system robust to disturbance light according to an embodiment of the present invention is determined according to the installation environment of the camera for photographing the user's face and the face recognition system, It may include a polarizing filter mounted on the front end.

According to an embodiment of the present invention, the face recognition system includes a receiving unit accommodating a plurality of polarizing filters, an external environment sensing unit detecting an external environment of the face recognition system, and at least one installed at the front end of the camera A filter mounting unit to which the above polarizing filter is mounted, and a filter selecting unit for selecting at least one of a plurality of polarizing filters in the receiving unit based on the external environment sensed by the external environment sensing unit and inserting it into the filter mounting unit and fixing it may include

According to an embodiment of the present invention, the external environment sensor is at least one illuminance sensor installed in the face recognition system, and the filter selector includes a plurality of polarized lights accommodated in the receiving unit based on the illuminance sensed by the illuminance sensor. At least one polarization filter may be selected among the filters.

According to an embodiment of the present invention, the face recognition system manages a table in which storage information for each illuminance is stored, and the filter selector selects storage information of a polarizing filter corresponding to the illuminance sensed by the illuminance sensor from the table. After searching, at least one polarization filter may be selected from the receiving unit based on the search result.

According to the above-described embodiments of the present invention, the installation place of the face recognition system is a place vulnerable to disturbance light, and if it is affected by this, the selective polarization filter according to time is automatically removable or can be installed at all times. By providing the system, it is possible to prevent a decrease in the face recognition rate due to the disturbance light.

1 is a diagram showing the configuration of a face recognition system robust to disturbance light according to an embodiment of the present invention.
2 to 4 are diagrams for explaining a structure and method of mounting an illuminance sensor of a face recognition system according to an embodiment of the present invention.

Hereinafter, specific embodiments of the present invention will be described with reference to the drawings. The following detailed description is provided to provide a comprehensive understanding of the methods, devices, and/or systems described herein. However, this is merely an example, and the present invention is not limited thereto.

In describing the embodiments of the present invention, if it is determined that a detailed description of a known technology related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted. In addition, the terms to be described later are terms defined in consideration of functions in the present invention, which may vary according to intentions or customs of users and operators. Therefore, the definition should be made based on the content throughout this specification. The terminology used in the detailed description is for the purpose of describing embodiments of the present invention only, and should in no way be limiting.

Hereinafter, a face recognition system according to an embodiment of the present invention will be described with reference to the accompanying drawings.

1 is a diagram illustrating the configuration of a face recognition system robust to disturbance light according to an embodiment of the present invention, and FIGS. 2 to 4 describe the structure and method of mounting an illuminance sensor of the face recognition system according to an embodiment of the present invention It is a drawing for

1 to 4, the face recognition system according to the embodiment of the present invention is A storage space in which a plurality of polarization filters are accommodated is provided, at least one or more of the plurality of polarization filters accommodated in the storage space is selected and inserted into the front end of the camera, and a table is stored in the face recognition system based on the experimental data. (Table: A table in which storage location information in which a polarizing filter according to illumination is set) detects the external environment (using the illumination sensor installed in the face recognition system), and then searches for storage location information in the table based on this , the polarization filter corresponding to the retrieved storage location information is pulled out from the storage unit and inserted and mounted on the front end of the camera, thereby minimizing the influence of disturbance light.

A polarizing film for manufacturing a polarizing filter to be applied to a face recognition system according to an embodiment of the present invention is prepared as an anti-reflective adhesive type. The basic structure of a polarizing film has a three-layer structure, with a polarizing element (PVA: Poly Vynyl Alcohol) positioned at the center, and a protective layer (TAC: Tri Acetyl Cellulose) laminated on the upper and lower surfaces. Then, an adhesive, a reflector, and a surface treatment are added thereon.

For the experiment, each of a circular polarization filter and a linear polarization filter are prepared and prepared. Circularly polarized light does not have directionality, but linearly polarized light has directionality.

Prepare a laboratory glass slide to which the film is to be attached. Glass slides are positively charged and uncoated, so choose a product with improved adhesion. If the thickness of the slide is not constant due to chromatic aberration of glass, errors may occur in experimental data due to errors due to chromatic aberration. Therefore, in order to reduce errors caused by chromatic aberration, the thickness may be set to 0.1 to 0.2 mm as a standard.

A polarizing film is attached to the slide. The slide is linearly polarized and circularly polarized, respectively.

It is attached according to the specifications of the circular polarizing filter, and 20 to 50 nanometers is selected for the circular polarizing filter and 20 to 50 nanometers is selected for the linear polarizing filter.

The main part in the experiment for the method of overcoming disturbance light using a polarizing filter is to prepare a polarizing filter suitable for the specifications and to prepare a lighting system that reproduces the disturbance light. As described in Section 2.3.1, the polarizing filter was prepared according to each specification.

The lighting system to reproduce the disturbance light was also described in Section 1.3, and the lighting conditions were set as the standard in Section 2.2.1. The composition of the laboratory is as follows.

■ Disturbance light reproduction illumination: 1, 2, 3, 4 / 4 light sources

■ Lab Base Lighting: Stroke Light

■ Face Recognition System: 1 pc.

■ Scale for distance measurement: 100 / 150 / 200

Figure 2-8 shows the laboratory and experimental scene configured according to the configuration of the laboratory.

As shown in Figure 2-8, the experimental method is to investigate the recognition rate by moving the face recognition camera away from the face recognition camera by 200cm and approaching it in 50cm units. Face recognition is performed three times under each condition, and the score of each run is recorded.

Therefore, the recognition experiment is performed three times without a polarizing filter, and then the recognition experiment is performed three times after attaching the polarizing filter. For the experiment, the disturbance light is reproduced, and the illuminance value for each disturbance light is measured and recorded. When the illuminance value is recorded, the experiment is carried out according to the following procedure.

Prior to such an experiment, illuminance measurement for each of 15 lighting conditions is preceded prior to conducting the experiment. The measured illuminance values are shown in the table below.

[Table 1]

A previous repeated experiment was carried out by installing an optical filter, but in the preceding experiment, the advantages and disadvantages of a linear polarization filter and a circular polarization filter were sought. In the case of a circular polarization filter, although it does not have directionality, it can be seen that the filtering effect is different according to the division between the front and the back because there is a division between the front and the back.

The linear polarization filter is divided into a horizontal filter and a vertical filter, and when these two filters are overlapped, it is possible to completely block the light by adjusting the angle. The advantage of the linear polarization filter is that it can completely block the disturbance light when the direction is set at right angles to the angle of incidence of the disturbance light. it's difficult. Therefore, it was found that if the angle of the two filters is finely adjusted by overlapping the two filters, the disturbance light blocking is effective, but the versatility cannot be guaranteed, and it is difficult to automatically execute the fine angle adjustment.

Therefore, in this experiment, the experiment of the linear polarization filter is postponed to the follow-up experiment, and the focus is on preparing a countermeasure to overcome the disturbance light through the circular polarization.

The laboratory accommodates the conditions of the conference room, sets the lighting conditions, and installs a face recognition system. Prepare a laptop computer to record the experimental data, and make sure it can be recorded immediately.

■ Experimenters: Park Ki-cheol / Yoon Kyung-yong / Nam Young-woo / Jang Jeong-hyeok

■ Face Recognition Experiment: Park Ki-cheol

■ Composition of lighting conditions / Removal of filters / Experimental progress: Kyung-Yong Yoon

■ Data Record Management: Jang Jeong-hyeok

■ Experimental equipment configuration support: Nam Young-woo

The experiment was conducted as shown in the upper photo of Figure 2-10. The upper right photo is a scene in which illuminance is measured according to lighting and data is recorded in real time accordingly, the lower left photo is a photo where face recognition failed due to excessive disturbance light, and the right photo is a circular polarization filter under the same conditions. It records the situation recognized when it is installed.

As shown in Figure 2-10, when the disturbance light is severe, the photoelectric sensor of the camera rises to the saturation voltage level by the reflected light, making recognition impossible. Therefore, by using a circular polarization filter to filter out the reflected and incident disturbance light, the recognition is possible.

Experimental data and analysis

The experimental data are shown in the table below.

The meaning of 100 N/F in Table 2-3 means that the distance is 100 cm (1 meter), No Filter, that is, no polarizing filter is installed. And 100 F/S means 100cm (1 meter), Filter Setup, that is, a polarizing filter is installed. Numbers represent recognition attempts and recognition scores. Since the recognition threshold is 65 points, a score less than 65 is considered a recognition failure.

Recognition experiments were performed 3 times each under the 15 lighting conditions in Table 2-2 at each distance and the filter installation condition. For example, 100 N/F illumination 1: 94, 92, 91 to 94, 92, 91 is a recognition score 3 times. Since all turns were less than 65 points, they were all cases of failure to recognize.

Recognition failed all 3 times under 100 N/F, that is, 15 times without a polarizing filter at a distance of 1 meter apart, but 3 times under the same lighting condition with a polarizing filter at 100 S/F and 1 meter apart. All were successful.

Under the conditions of 150 N/F, that is, there is no polarizing filter at a distance of 1.5 meters, and under illumination conditions 10, 12, 13, and 15, recognition of each condition failed. In the mounting conditions, recognition was successful in all lighting conditions.

At 200 N/F, that is, there is no polarizing filter at a distance of 2 meters, and under illumination conditions 10, 12, 13, and 15, recognition of each condition failed, but at 200 F/S, at a distance of 2 meters, polarizing filter In the mounting conditions, recognition was successful in all lighting conditions.

Table 2-3 data are shown in Figure 2-11, Figure 2-12, and Figure 2-13, respectively. In Figures 2-11, 12, and 13, the black diamond indicates the N/F condition, and Solid, Half-solid, and Empty indicate the 1st, 2nd, and 3rd recognition attempts. The red circle is the F/S condition, and Solid, Half-solid, Empty means the first, second, and third recognition attempts.

Experiment conclusion

As shown in Figure 2-11, it can be seen that the recognition rate is better when the polarization filter is not installed in most lighting conditions at a short distance within 1 meter. However, under No. 15 lighting condition, that is, left 1,2 (Light 1-1, Light 1-2), Right 1,2 (Light 2-1, Light 2-2) are all ON, the polarizing filter may be installed. was recognized when

Separately, it can also be seen that the greater the distance from the recognition device, that is, the greater the distance from the recognition device is, the higher the recognition rate score is when the polarizing filter is installed.

This fact is that the camera's illuminance sensor default value condition is set to the center of the short distance, and as the distance increases, the digital iris value slows down, so it seems to accept a lot of incident light. Therefore, it can be seen that as the distance from the recognition device increases, the brightness of the photographed image is bright and recognition is difficult.

Comprehensively judging the experimental results, it seems that the installation site of the face recognition system is vulnerable to disturbance light, and if it is affected by this, it is safe to perform selective automatic removal and installation by time or to always install the face recognition system. If the system has a frontal distance of more than 4 meters from a subject, i.e., a person, and wants a walk-through type of recognition, i.e. remote recognition, it is better not to install a polarizing filter. However, if there is a lot of disturbing light nonetheless, selective automatic removal and installation by time is recommended.

The face recognition system according to an embodiment of the present invention for detecting disturbance light may be equipped with a plurality of illuminance sensors as shown in the figure below.

Specifically, by implementing a plurality of illuminance environments, it was found that the disturbance light reflected to the lower part and incident had the greatest effect on the recognition system as a result of the experiment. Because of this, the fact that the error rate was high was taken into account.

Therefore, the incident angle was generally found to be the most vulnerable to the lower incident disturbance light, and the direction was vulnerable to the incident light at the rear side, but particularly to the disturbance light incident at 90 degrees from the rear side. Accordingly, it may be desirable to attach the sensor as shown in the figure above.

The front upper sensor is a constant sensor, so fix it, reserve two terminals for plug-in in the upper and lower parts of the rear part, 3 terminals each in the side part, and 2 terminals in the lower part. do.

The sensor enables installation only with plug-in without opening the system or perforating the housing, and obtains and applies the value according to the position of the sensor in the program.

The foregoing description of the present application is for illustration, and those of ordinary skill in the art to which the present application pertains will understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present application. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. For example, each component described as a single type may be implemented in a dispersed form, and likewise components described as distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the following claims rather than the above detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present application. .

50 : camera
100: face recognition system
110: storage unit
112: polarizing filter
120: filter mounting part
122: filter selection unit
124: memory
130: light sensor

Claims (1)

A face recognition system for recognizing a face, comprising:
A camera for photographing the user's face;
and at least one polarizing filter determined according to the illumination environment of the face recognition system and mounted on the front end of the camera.

Images