KR101668555B1 - Method and apparatus for recognizing worker in working site image data - Google Patents

Abstract

Regardless of the circumstances such as when there is a feature change due to the worker's state change, when there are a large number of workers, when the worker's size changes, And provides a technique capable of clearly recognizing the above workers from the photographed image. A method for recognizing a worker in a worksite image data according to an embodiment of the present invention is a method for recognizing a worker in an image data of a predetermined area by using at least one object detected by using an outline constituting the object, Selecting an object matching the shape of the reference object as a region of interest; Determining whether the color information included in the ROI is matched with reference color information existing in the database using a predetermined algorithm; And selecting a region of interest including the color information determined to match with the reference color information existing in the database as an operator object.

Description

METHOD AND APPARATUS FOR RECOGNIZING WORKER IN WORKING SITE IMAGE DATA [0002]

TECHNICAL FIELD The present invention relates to a technology for recognizing an operator who performs work in a work site such as a construction site or a construction site. More specifically, the present invention relates to a technique for recognizing a plurality of workers in real time at a work site, And to secure the stability of the work site by accurately grasping the work area.

According to statistics for construction industry deaths in 2012, deaths related to construction equipment operation in construction sites accounted for more than 25% of the total. In addition, the number of deaths related to the operation of construction equipment in construction sites is 91 in 2012, an increase of more than 38% compared to 66 in 2008. According to statistics from the Ministry of Employment and Labor, the economic loss incurred by industrial disasters in the construction industry in 2012 accounted for about 33% of the total industrial disaster cost. As a result of estimating the ratio of the number of deaths and accidents (91 persons) related to the construction equipment operation among the deaths caused by the accident at the construction site in 2012 (357 persons), the economic loss cost It is estimated to reach about KRW 1.6 trillion (Korea Occupational Safety and Health Agency, 2012). As this trend is expected to increase every year, there has been a need to develop technology for securing construction equipment safety to prevent safety accidents related to construction equipment operation in the construction site. According to the Occupational Safety and Health Administration (OSHA) Construction Safety Accident Statistics, 293 serious injuries have been caused by industrial trucks. The cause of this was occupational environment related factors such as visual field disorder or collision with load and other workers (57.7%) (Korea Occupational Safety and Health Agency, 2008). In order to prevent safety accidents related to operation of construction equipment in the construction site caused by factors related to the work environment, it is necessary to develop a technology capable of detecting real-time risk factors covering all aspects of the construction equipment working environment.

Accordingly, a number of techniques have been proposed that can detect workers in real-time among the risk factors in the construction equipment working environment. The proposed method is based on recognition of an operator by using a distance sensor such as an ultrasonic sensor, a detection sensor, a time of arrival (ToA) method, a radio frequency identification (RFID) technology, a Pyroelectric Infrared Ray (PIR) And a method for recognizing an operator using an image sensor.

In the "Heavy equipment worker approach detection device (application number: 10-1997-0066678)" and "proximity alarm generation system (application number: 10-2001-0014181)", the ultrasonic sensor is used to recognize the worker in the construction equipment workspace Respectively.

In the "Construction Equipment Safety Alarm System (Application No .: 20-2005-0000816)", the sensor was used to recognize workers in the construction equipment workspace.

In the "Tunnel Construction Site Equipment - Personnel Safety Management System (Application No .: 10-2011-0031713)" and "Construction Site Safety Accident Prevention System (10-2011-0065579)", a ToA (Time of Arrival) And to recognize workers in construction equipment workspace.

In the "Safety management device using RFID and its method (Application No. 10-2012-0105578)", RFID (Radio Frequency Identification) technology was used to recognize workers in a construction equipment workspace.

In the "Heavy Equipment Access Monitoring System for Construction (Application No. 10-2013-0033357)", a PIR (Pyroelectric Infrared Ray) sensor and an ultrasonic sensor were used to recognize workers in a construction equipment workspace.

"A study on the development of technology to detect all obstacles in the excavator working area (Sojyun et al., 2010)" tried to recognize a worker in a construction equipment workspace by using a two-dimensional laser scanner.

In the case of a method using an ultrasonic sensor, a detection sensor, and a Pyroelectric Infrared Ray (PIR) sensor, the detection distance is shorter than a few meters to detect an operator using the distance sensor, About 12m), there is a disadvantage that it is impossible to recognize the worker.

In the case of the method using the RFID (Radio Frequency Identification) technology, the detection distance is sufficient but the tag is attached to all construction equipment and workers in the construction site. In the case of the method using the two-dimensional laser scanner, although the detection distance is sufficient, there is a disadvantage that the sensor mounting cost ranges from several million won to tens of thousands of won.

On the other hand, the video image obtained by using the image sensor from the construction site has an advantage that it can be utilized for the recognition of the worker within the working radius of the construction equipment (about 12 m) at a relatively low cost.

In this paper, we propose a method to extract the moving pixels from the video images obtained from the construction site using the image sensor, To identify the worker.

"An Object Recognition, Tracking, and Contextual Reasoning-based Video Interpretation Method for Rapid Productivity Analysis of Construction Operations (Gong and Caldas, 2011)" and "Worker Detection in Video Frames for Initializing Vision Trackers (Park and Brilakis 2012) In this study, we extract motion picture pixels from video images obtained by using image sensor from a construction site, and recognize the operator by using the colors of the safety vest, which is a type of worker 's type and safety protection equipment, as characteristics of the worker.

In this paper, we propose a method to extract the moving image from the video image obtained from the construction site using the image sensor, The color of the safety helmet and safety vest, which is a kind of protective gear, and the position and ratio of safety helmet and safety vest in the line condition, were used to recognize the operator.

"Vision Watching System and Method for Safety Hat (Patent No .: US 8,279,277 B2)" We extract motion picture pixels from a video image obtained from a construction site using an image sensor, And to recognize the worker by utilizing it as a characteristic of the worker.

A common disadvantage of these methods is that they can not be perceived by workers who are standing, standing or sitting. In addition to these fundamental drawbacks, Lee and Hong (2014) have a drawback in that only workers wearing safety helmets and safety vests at the same time can recognize them.

In a study on "Construction Safety Visualization (Shrestha et al., 2011)", the shape of the safety helmet, which is a type of safety protection equipment, is used as a characteristic of the worker from the video image obtained by using the image sensor from the construction site. To identify the worker. The disadvantage of this method is that the background can be applied only to simple video images, and there is a disadvantage that it can not be recognized when a large number of workers are working in the video image. In addition, when the user is leaning or leaning in the opposite direction to the image sensor, there is a problem that the safety helmet is obstructed in the field of view, which is a disadvantage in that the operator can not be perceived.

Therefore, the present invention can be applied to a situation in which a worker operates in a stationary state, a characteristic change occurs in accordance with a worker's state change, there are a plurality of workers, The present invention has an object to provide a technique capable of clearly recognizing at least one operator from a photographed image.

According to an embodiment of the present invention, there is provided a method for recognizing a worker in worksite image data, the method comprising: extracting at least one object detected using an outline constituting an object from image data of a predetermined area, Selecting an object matching with a type of a reference object existing in a previously stored database as a region of interest; Determining whether color information included in the ROI is matched with reference color information existing in the database using a predetermined algorithm; And selecting the region of interest including the color information determined as matching with the reference color information existing in the database as an operator object.

Meanwhile, among the at least one objects detected by using the outline constituting the object from the image data photographed in the preset area, the worker recognizing device in the worksite image data according to the embodiment of the present invention may include a pre- A first determination unit that selects an object matching with the type of the reference object existing in the ROI as the ROI; A second determination unit for determining whether the color information included in the ROI is matched with reference color information existing in the database by using a predetermined algorithm; And a recognition result derivation unit for selecting a region object of interest including color information determined to match with reference color information existing in the database as an operator object.

According to the present invention, the object corresponding to the worker is firstly recognized through the outline of the work type of the worker, and then the color of the safety protective equipment worn by the worker is compared with the consideration of the influence of the external environment and the shooting angle, The operator is finally recognized.

Accordingly, as the above-mentioned element, it is possible to recognize the operator even in the case where the worker is stationary and recognize the operator regardless of the shape of the worker and the change in the image size according to the photographing distance. That is, it is possible to recognize the operator even in a photographing direction in which the operator's helmet is not visible, and it is possible to recognize even when a large number of operators are working in the image.

By virtue of the above-described effects, it is possible to secure the stability of the worksite by recognizing the worker at a job site with a very accurate probability compared with existing techniques.

1 and 2 are flowcharts of a worker recognition method in worksite image data according to an embodiment of the present invention.
3 is a block diagram of a worker recognizing apparatus in worksite image data according to an embodiment of the present invention.
Figure 4 is an example of a reference object image stored in a database for implementation of an embodiment of the present invention.
5 to 7 are examples of reference color information and background color information stored in a database for implementation of an embodiment of the present invention.
8 is an experimental example of recognizing an operator in an image according to an embodiment of the present invention.

Hereinafter, a method and apparatus for recognizing a worker in worksite image data according to an embodiment of the present invention will be described with reference to the accompanying drawings.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. The following examples are intended to illustrate the present invention and should not be construed as limiting the scope of the present invention. Accordingly, equivalent inventions performing the same functions as the present invention are also within the scope of the present invention.

In the following description, the same reference numerals denote the same components, and unnecessary redundant explanations and descriptions of known technologies will be omitted.

In the embodiments of the present invention, " communication ", " communication network ", and " network " The three terms refer to wired and wireless local area and wide area data transmission and reception networks capable of transmitting and receiving a file between a user terminal, a terminal of another user, and a download server.

1 and 2 are flowcharts of a worker recognition method in worksite image data according to an embodiment of the present invention. In the following description, a method of recognizing a worker in the worksite image data according to an embodiment of the present invention is performed by each configuration of a worker recognition apparatus in worksite image data according to an embodiment of the present invention But will not be limited thereto.

Referring to FIG. 1, first, from among image data obtained by photographing a predetermined region on a work site, among a plurality of objects detected using an outline constituting an object, (S10) of selecting at least one object matched with the shape of the ROI as an ROI object.

For example, in order to improve the performance of step S10, the apparatus can apply a process for recognizing a region of interest (a person recognized as an operator) in the entire video image based on Fast Feature Pyramids to photographed image data have.

At this time, when selecting the region of interest, the shape of the reference object existing in the database stored in the device and the shape of the recognized objects can be compared. That is, the Fast Feature Pyramids method is applied based on all types of databases that the operator can take, and the candidate objects that can be the interested region object, that is, the worker, are firstly selected. In Fig. 4, an example of data stored in such a database is shown.

4 is an example of a reference object image stored in a database for implementation of an embodiment of the present invention.

Referring to FIG. 4, the database includes a plurality of pieces of type information 100, which are information indicating the type of the worker corresponding to a line state, a walking state, a sitting state, a leaning state, a sitting state, a standing state, The image data as an example is shown. It goes without saying that only the outline data other than the above video data can be stored in accordance with the database.

In each of the plurality of type information 100, detailed form information 110 that can be changed according to the operator can be stored. The detailed shape information 110 may store image information corresponding to a shape that can be changed according to a direction or the like.

The plurality of type information 100 and the detailed type information 110 are utilized as a reference object and used as a criterion for selecting a region of interest when performing the step S10.

As shown in FIG. 4, various reference object information can be stored in the database, and it is possible to update the information of the learning module of the database, that is, various reference objects continuously, You can do it.

On the other hand, in step S10, in order to recognize the object irrespective of the size change of the shape according to the photographing distance, the apparatus may compare the contour of the object with the contour of the object while changing the size of the reference object.

The apparatus then determines whether the color information included in the ROI is matched with the reference color information existing in the database by using a predetermined algorithm (S20).

In order to recognize the worker from the image data at the work site, it is not sufficient to carry out the step S10 of accurately recognizing the worker based on the shape or the outline. That is, if the helm is not recognized by the direction or the form of the worker, or the outline form of the background such as the land or the apparatus is not similar to the worker, This is because an error in recognition of the operator may occur only by performing the step.

For this purpose, the apparatus uses the color included in the objects as a characteristic of the worker in the secondary recognition process in step S20 for the recognized area objects recognized through the first recognition process in step S10, It is possible to ensure the accuracy of the recognition by discriminating the error that may occur as a result of the execution.

That is, after performing step S20, the apparatus finally performs a step S30 of selecting an ROI including the color information determined to match with the reference color information existing in the database as an operator object, It is possible to obtain the effect of completely solving the problem of recognition error of the technology.

As described above, the reference color information is color information that can be utilized as a characteristic of an operator, that is, information stored in a database for selecting a worker object. The reference color information includes a safety protection gear Is color information corresponding to the color of the color image. Safety protective gear may include, for example, a safety vest, which is an accessory that the operator generally wears at the job site, in addition to a safety helmet.

FIG. 2 shows an example of a flow for determining whether the reference color information and the color information included in the ROI are matched using the preset algorithm.

Referring to FIG. 2, the apparatus firstly stores information on the reference color information, that is, the color space of the safety protection device and the reference color information including the component, (S21) from the database.

Thereafter, a color model-based pixel classification method is applied to the color information included in the ROI object and the loaded reference color information to determine whether the color information included in the ROI object matches the reference color information Specifically, this process is as follows.

That is, a step of performing a color space conversion operation on the color information included in the ROI, and extracting the color space and the color component (information having the same concept as the component of S21) from the color information included in the ROI S22) is performed first. Then, the color space in which the distribution of the color and the background color of the safety protection equipment included in the loaded reference color information is distinguished, and the color space and the color component information extracted from the component, (S23) of judging whether or not the color information included in the ROI matches the reference color information is performed.

In the secondary recognition process represented by step S20, as described above, the safety protection device (safety helmet or safety vest) is characterized by detecting an error that may occur in the recognition process of step S10 of FIG. 1 as described above, In order to recognize the operator, the image processing technique is applied based on the color invariant characteristics of the safety protective gear.

Image data is represented by a space of RGB (Red, Green, Blue) color, but it is susceptible to changes in external environment such as light. Accordingly, as described in the steps S21 to S23, the color space conversion operation is required to correctly recognize the safety protection gear using the color information.

Among the various color spaces, for example, the color spaces such as Normalized rgb, HSI (Hue, Saturation, Intensity), CIELUV, and YCbCr are hardly affected by changes in the external environment such as light, Has come. In each color space, the safety protective gear has a certain color distribution by the three components constituting the color space. In order to recognize the effective safety protection gear, it is necessary to select the color space and the component which clearly distinguish the color distribution and the background color distribution of the safety protection gear (corresponding to S21).

In the present invention, for example, for each of the 45 components in the fifteen color spaces, the three safety hat colors, the safety vest and the background color distribution are clearly distinguished from each other through factor selection. Based on the selected components, a color model-based pixel classification method, which is one of the image segmentation techniques, is applied to recognize the respective safety protection gears (steps S22 and S23). The color model-based pixel classification method has an advantage that the object can be recognized quickly and effectively based on the color invariant feature of the object through color model learning.

Examples of reference color information and background color information that can be stored in the database for performing steps S21 to S23 are shown in FIGS.

Referring to FIG. 5, the color information 200 can be identified based on each color of the helmet of various colors (white, yellow, and red). The color distribution information 210 may be separately stored in each color information so that the color space and corresponding components can be selected.

Referring to FIG. 6, the safety vest color distribution information 310 stored in the database is shown, and it is confirmed that the safety vest 310 is provided to be able to select a color space and corresponding components with respect to the safety vest .

In FIG. 7, it can be seen that color information about soil and other background elements is stored as a background. The color information 400 is stored according to the type of each background and similarly the color distribution information 410 is divided and stored so that the color space and components included in the reference color information of FIGS. .

An experimental example of recognizing the operator through such a process is shown in Fig.

Referring to FIG. 8, there is shown an experimental example of a result of recognition of an operator by the above-described technique of the present invention in the image data 500. First, the objects 510 and 520 indicate the ROI objects selected through step S10 of FIG. On the other hand, the bright object 511 is used as the color information used to finally select the worker object by performing step S20 of FIG. 1 and steps S21 to S23 of FIG. 2, that is, based on the safety protection gear color database The results are recognized by applying the color model method.

As shown in FIG. 8, an object 510 including an object 511 recognized as a bright portion is an object finally recognized as an operator, and the other object 520 is selected as a region of interest, It means an object not recognized as an operator.

In other words, according to the present invention, it is possible to completely solve the problem of the existing recognition process, recognize the operator more accurately, and increase the stability of the worksite through the recognition process over two steps Can be confirmed.

3 is a block diagram of a worker recognizing apparatus in worksite image data according to an embodiment of the present invention. In the following description, portions overlapping with the description of Figs. 1, 2 and 4 to 8 will be omitted.

3, the worker recognizing apparatus 10 in the worksite image data according to an embodiment of the present invention includes a first determining unit 11, a second determining unit 12, and a recognition result deriving unit 13 And the first determination unit 11 and the second determination unit 12 are connected to the database 40 to receive reference information.

1, the first determination unit 11 performs a function of receiving the image data of the photographed region 30 from the photographing apparatus 20, And selects a region of interest using the shape of the reference object received from the user. In the present invention, the photographing apparatus 20 includes all the photographing apparatuses having mobility installed in the work equipment together with the CCTV and other photographing apparatuses installed at the work site.

The second determining unit 12 is a unit for performing the functions of steps S21 to S23 in Fig. 1 and step S21 to S23 in Fig. 2. The second determining unit 12 obtains the reference color information from the database 40 from the first determining unit 11, And performs a function of comparing the object information and the color information included therein with specific functions of S20 to S23.

The recognition result deriving unit 13 performs the function of step S30 in Fig. 1, and performs the function of finally recognizing the operator in the image data according to the determination result of the second determination unit 12. Fig.

As described above, the worker identification method in the worksite image data according to the embodiment of the present invention can be implemented by an application installed in the terminal (which may include a program basically installed in the terminal or a program included in the operating system) And may be executed by an application (that is, a program) directly installed on the terminal by a user via an application providing server such as an application store server, an application, or a web server related to the service. In this sense, the worker recognition method in the worksite image data according to the above-described embodiment of the present invention is basically installed in the terminal or implemented as an application (i.e., program) directly installed by the user, And recorded on a recording medium.

Such a program may be recorded on a recording medium that can be read by a computer and executed by a computer so that the above-described functions can be executed.

In order to execute the worker recognition method in the worksite image data according to the embodiments of the present invention, the above-described program may be stored in a computer language such as C, C ++, JAVA, And may include coded codes.

The code may include a function code related to a function or the like that defines the functions described above and may include an execution procedure related control code necessary for the processor of the computer to execute the functions described above according to a predetermined procedure.

In addition, such code may further include memory reference related code as to what additional information or media needed to cause the processor of the computer to execute the aforementioned functions should be referenced at any location (address) of the internal or external memory of the computer .

In addition, when a processor of a computer needs to communicate with any other computer or server, etc., to perform the above-described functions, the code may be stored in a computer's communication module (e.g., a wired and / ) May be used to further include communication related codes such as how to communicate with any other computer or server in the remote, and what information or media should be transmitted or received during communication.

The functional program for implementing the present invention and the related code and code segment may be implemented by programmers in the technical field of the present invention in consideration of the system environment of the computer that reads the recording medium and executes the program, Or may be easily modified or modified by the user.

Examples of the computer-readable recording medium on which the above-described program is recorded include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical media storage, and the like.

Also, the computer-readable recording medium on which the above-described program is recorded may be distributed to a computer system connected via a network so that computer-readable codes can be stored and executed in a distributed manner. In this case, any of at least one of the plurality of distributed computers may execute some of the functions presented above and transmit the result of the execution to at least one of the other distributed computers, and transmit the result The receiving computer may also perform some of the functions described above and provide the results to other distributed computers as well.

In particular, a computer-readable recording medium on which an application, which is a program for executing a worker recognition method in job site image data according to each embodiment of the present invention, is recorded, includes an application store server, (E.g., a hard disk) included in an application provider server such as a related web server, or an application providing server itself.

A computer capable of reading a recording medium on which an application, which is a program for executing a worker recognition method in the worksite image data according to each embodiment of the present invention, can be read is not limited to a general PC such as a general desktop or a notebook computer, A mobile terminal such as a personal computer, a personal digital assistant (PDA), and a mobile communication terminal. In addition, the present invention should be interpreted as all devices capable of computing.

In addition, a computer capable of reading a recording medium on which an application, which is a program for executing a worker recognition method in the worksite image data according to an embodiment of the present invention, is read is a smart phone, a tablet PC, a PDA (Personal Digital Assistants) , The application may be downloaded from the application providing server to the general PC and installed in the mobile terminal through the synchronization program.

While the present invention has been described in connection with what is presently considered to be the most practical and preferred embodiments, it is to be understood that the invention is not limited to the disclosed embodiments. That is, within the scope of the present invention, all of the components may be selectively coupled to at least one. In addition, although all of the components may be implemented as one independent hardware, some or all of the components may be selectively combined to perform a part or all of the functions in one or a plurality of hardware. As shown in FIG. The codes and code segments constituting the computer program may be easily deduced by those skilled in the art. Such a computer program can be stored in a computer-readable storage medium, readable and executed by a computer, thereby realizing an embodiment of the present invention. As a storage medium of the computer program, a magnetic recording medium, an optical recording medium, or the like can be included.

It is also to be understood that the terms such as " comprises, "" comprising," or "having ", as used herein, mean that a component can be implanted unless specifically stated to the contrary. But should be construed as including other elements. All terms, including technical and scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs, unless otherwise defined. Commonly used terms, such as predefined terms, should be interpreted to be consistent with the contextual meanings of the related art, and are not to be construed as ideal or overly formal, unless expressly defined to the contrary.

The foregoing description is merely illustrative of the technical idea of the present invention, and various changes and modifications may be made by those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are intended to illustrate rather than limit the scope of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be construed according to the following claims, and all technical ideas within the scope of equivalents should be construed as falling within the scope of the present invention.

Claims (10)

Selecting an object matching the shape of a reference object existing in a pre-stored database among the at least one objects detected using the outline constituting the object, from the image data of the predetermined area; Determining whether color information included in the ROI is matched with reference color information existing in the database using a predetermined algorithm; And selecting a region of interest, which includes color information determined to match with reference color information existing in the database, as an operator object,
Wherein the step of determining, using the predetermined algorithm,
Loading the reference color information including the selected color space and the component from the database so that the color distribution and the background color distribution of the safety protection gear are distinguished; And applying color model-based pixel classification to the color information included in the ROI and the loaded reference color information to determine whether the color information included in the ROI matches the reference color information Comprising:
The method of claim 1,
Extracting color space and color component information from color information included in the ROI by performing a color space conversion operation on the color information included in the ROI; And a color space and a component in which a distribution of the color distribution and a background color of the safety protective gear included in the loaded reference color information are distinguished, and the extracted color space and color component information, And determining whether or not the color information included in the ROI matches the reference color information based on the comparison result. The method according to claim 1,
The reference color information may include:
Wherein the color information corresponding to the color of the safety protective gear worn by the operator is information stored for selection by the operator object. delete delete The method according to claim 1,
The method of claim 1,
The region of interest is selected by comparing the size of the reference object with the contour of the object in order to select an object having an outline that matches the shape of the reference object irrespective of the size of the object, Wherein the worker recognizes the worker in the worksite image data. A first matching unit which selects an object matching the shape of a reference object existing in a previously stored database among at least one of the objects detected by using the contour constituting the object from the image data of the predetermined area, A determination unit; A second determination unit for determining whether the color information included in the ROI is matched with reference color information existing in the database by using a predetermined algorithm; And
And a recognition result derivation unit for selecting a region of interest, which includes color information determined to match with reference color information existing in the database, as an operator object,
The second determination unit may load the reference color information selected from the color space and the component that distinguishes the color distribution and the background color distribution of the safety protection equipment from the database, Determining whether or not color information included in the ROI is matched with the reference color information by applying a color model based pixel classification method to the loaded reference color information,
The second determination unit may determine,
A color space conversion process is performed on the color information included in the ROI to extract color space and color component information from the color information included in the ROI, The color space and the component in which the distribution of the color and the background color of the protective gear are distinguished are compared with the extracted color space and color component information to determine whether the color information included in the ROI matches the reference color information And a determination unit operable to determine whether the worker is in the on-scene video data. The method according to claim 6,
The reference color information may include:
Wherein the color information corresponding to the color of the safety protective gear worn by the operator is information stored for selection by the operator object. delete The method according to claim 6,
The first determination unit may determine,
The region of interest is selected by comparing the size of the reference object with the contour of the object in order to select an object having an outline that matches the shape of the reference object irrespective of the size of the object, And the worker recognizing device in the worksite image data. Selecting an object matching the shape of a reference object existing in a pre-stored database among the at least one objects detected using the outline constituting the object, from the image data of the predetermined area; Determining whether color information included in the ROI is matched with reference color information existing in the database using a predetermined algorithm; And selecting a region of interest, which includes color information determined to match with reference color information existing in the database, as an operator object,
Wherein the step of determining, using the predetermined algorithm,
Loading the reference color information including the selected color space and the component from the database so that the color distribution and the background color distribution of the safety protection gear are distinguished; And applying color model-based pixel classification to the color information included in the ROI and the loaded reference color information to determine whether the color information included in the ROI matches the reference color information Comprising:
The method of claim 1,
Extracting color space and color component information from color information included in the ROI by performing a color space conversion operation on the color information included in the ROI; And a color space and a component in which a distribution of the color distribution and a background color of the safety protective gear included in the loaded reference color information are distinguished, and the extracted color space and color component information, And determining whether or not the color information included in the ROI matches the reference color information. The method of claim 1, further comprising: Readable recording medium.

Images