KR20230157856A - Machine vision camera based on deep learning - Google Patents

Abstract

The present invention is composed of individual modules and includes a module part for selecting and assembling each module, wherein the module part is a lens mount part where a lens is assembled and can be assembled according to the specifications of the lens, and a wavelength or brightness control unit. a lighting unit, a sensor unit that receives light incident upon the lens passing through the lens mount unit, an FPGA unit that verifies image processing operation and performance, an embedded unit that analyzes the image and receives the results, and the embedded It relates to a machine vision camera based on deep learning that includes an interface unit that sends output signals to the outside of the results analyzed in the unit.

Description

Machine vision camera based on deep learning}

The present invention relates to a machine vision camera based on deep learning. It is a deep learning-based machine vision camera that can produce an optimal smart camera by combining each module in a modular manner to suit the usage environment. This is about a machine vision camera based on learning.

In general, with the advancement of computer technology, the use of deep learning technology and machine vision functions based on it is increasing in various fields.

Among various fields, looking at the camera field, with the development of digital technology, digital cameras are becoming widely used, replacing conventional film cameras.

In particular, DSLRs, which are digital conversions of SLR (single-lens reflex) cameras, which were sold at high prices in the past and were known to be used only by professionals, are becoming popular relatively inexpensively.

DSLR cameras, also called 'interchangeable lens cameras', are an abbreviation for digital single-lens reflex cameras. Unlike typical digital cameras that have a double lens, DSLR cameras take pictures by sending images to the viewfinder through a single lens. It refers to a single-lens reflex camera.

A DSLR is equipped with a mirror that can be rotated within a predetermined angle on the optical axis of the lens, so that light passing through the lens is reflected from the mirror and is viewed by the photographer through a pentaprism and viewfinder. You can check it.

At this time, when a shutter-release signal is input, the mirror is rotated around its axis and raised to deviate from the optical axis of the lens, and when the shutter is opened, an image of the subject is formed on the imaging device.

DSLR has much better image quality and color than regular digital cameras, so it is gaining popularity among users who want to obtain clearer images.

In general, a DSLR consists of a camera body (body) and separate, attachable and detachable interchangeable lenses.

When the interchangeable lens is combined with the body, power is supplied from the main body to the lens, and electrical signals are transmitted and received between the lens and the body.

Like a typical digital camera, the body of a DSLR camera includes a shutter-release button, a mode dial for setting the shooting mode, a wide-angle-zoom button or a telephoto-zoom button, and a display unit such as a function button and an LCD window to indicate the subject to be shot and the composition. It is equipped with a viewfinder, a small window for setting settings.

However, in the case of a DSLR camera, various interchangeable lenses are mounted on the body as needed to capture images. The body of a DSLR camera is not only much heavier than that of a regular digital camera, but is also sold at a relatively high price.

Therefore, in order to produce a good DSLR-grade image, there was a problem in that one had to purchase a body as well as an interchangeable lens.

In other words, the actual usage environment of manufacturers who need smart cameras is very diverse, and the selection of sensors with resolution suitable for each situation, selection of embedded devices according to the difficulty of image processing, determination of interface for transmitting output values to the outside, and lighting There are many things to consider, such as selection of wavelength or brightness, and selection of lens magnification and depth of field.

Additionally, the types of smart cameras on the market are extremely limited, and in particular, all hardware components that make up a smart camera, including sensors and lights, are fixed.

Therefore, not only was it difficult to select a smart camera suitable for the actual manufacturing environment, but it was also difficult to change some of the configuration of the smart camera.

These problems often necessitate the development of new models, but in reality, it is very difficult for manufacturers to respond to all of these cases.

Therefore, there is a problem in that small and medium-sized manufacturers can be extremely limited in forming a smart factory by applying machine vision for production or quality inspection.

Patent Document 1: Republic of Korea Patent Publication No. 10-2005-00869976 (Publication Date: 2005.08.30)

The present invention is intended to solve the above-mentioned problems. It is a deep learning-based machine vision camera that can produce an optimal smart camera by combining each module according to the usage environment in each module method. The purpose is to provide a machine vision camera that

The problems to be solved by the present invention are not limited to the problems mentioned above, and other problems not mentioned can be clearly understood by those skilled in the art from the description below.

In order to solve the above-mentioned problem, it is composed of each module according to the present invention, and includes a module part for selecting and assembling each module,

The module part includes a lens mount part on which a lens is assembled and can be assembled according to the specifications of the lens, a lighting part that adjusts the wavelength or brightness, and a sensor part that receives light incident on the lens through the lens mount part. It includes an FPGA unit that verifies image processing operations and performance, an embedded unit that analyzes the image and receives the results, and an interface unit that sends an output signal to the outside of the results analyzed by the embedded unit.

It may further include a rear cover formed according to the shape of the interface.

The module part may include a mechanism configured to enable assembly.

The inner walls of the sensor unit, FPGA unit, embedded unit, and interface unit may be composed of a PCB BOARD.

A module unit composed of each module and selecting and assembling each module; It includes a coupling part that combines the module part, wherein the module part includes a lens mount part on which a lens is assembled and can be assembled according to the specifications of the lens, a lighting part that adjusts the wavelength or brightness, and the lens is mounted in the lens mount part. A sensor unit that receives incident light, an FPGA unit that verifies image processing operation and performance, an embedded unit that analyzes the image and receives the results, and outputs the results analyzed by the embedded unit to the outside. It includes an interface unit that sends the sensor, and the coupling unit includes a rear cover unit that is coupled to the sensor unit.

As such, the present invention is a machine vision camera based on deep learning, and has the effect of producing an optimal smart camera by combining each module in a modular manner to suit the usage environment.

In other words, more than hundreds of combinations can be created, which has the effect of providing smart cameras suitable for various manufacturing environments.

In addition, even if assembly has already been completed and installed in the field, it has the effect of being easily replaced when one or multiple modules need to be changed.

This has the effect of enabling manufacturers to produce customized smart cameras that meet the diverse needs of customers.

The effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

1 to 3 are diagrams showing a machine vision camera based on deep learning according to an embodiment of the present invention.
Figure 4 is a diagram showing the state of combining a machine vision camera and lens based on deep learning of Figure 1.
Figure 5 is a diagram showing a state in which a lens is coupled to the deep learning-based machine vision camera of Figure 4.
FIG. 6 is a diagram showing the disassembled state of the deep learning-based machine vision camera of FIG. 1.
Figures 7 and 8 are diagrams showing a machine vision camera based on deep learning according to another embodiment of the present invention.

Since the present invention can make various changes and have various embodiments, specific embodiments will be illustrated in the drawings and described in detail.

However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all changes, equivalents, and substitutes included in the spirit and technical scope of the present invention. While describing each drawing, similar reference numerals are used for similar components.

When a component is said to be "connected" or "connected" to another component, it is understood that it may be directly connected to or connected to the other component, but that other components may exist in between. It should be. On the other hand, when it is mentioned that a component is “directly connected” or “directly connected” to another component, it should be understood that there are no other components in between.

The terms used in this application are only used to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this application, terms such as “comprise” or “have” are intended to designate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are not intended to indicate the presence of one or more other features. It should be understood that this does not exclude in advance the possibility of the existence or addition of elements, numbers, steps, operations, components, parts, or combinations thereof.

Hereinafter, embodiments of the present invention will be described with reference to the attached drawings. The same symbols shown in each drawing represent the same members. In describing the present invention, detailed descriptions of related known functions or configurations are omitted in order to not obscure the gist of the present invention.

Figures 1 to 3 are diagrams showing a machine vision camera based on deep learning according to an embodiment of the present invention, and Figure 4 shows a state of combining the machine vision camera and lens based on deep learning of Figure 1. Figure 5 is a diagram showing a state in which a lens is combined in the machine vision camera based on deep learning in Figure 4, and Figure 6 is a diagram showing the disassembled state of the machine vision camera based on deep learning in Figure 1. This is the drawing shown.

Referring to FIG. 1, a machine vision camera 1 based on deep learning includes a module unit 10.

The module unit 10 is composed of individual modules, and each module can be selectively assembled to desired specifications according to the environment and situation.

For this purpose, the module unit 10 includes a lens mount unit 11, a lighting unit 12, a sensor unit 13, an FPGA unit 14, an embedded unit 15, and an interface unit 16. It can be included.

Here, the lens mount unit 11 is assembled with a lens, and assembly may be possible depending on the specifications of the lens 11a.

Additionally, the lens mount unit 11 can be selectively used with various types of lenses 11a.

Additionally, the lens mount unit 11 may include a mounting unit on which the lens 11a is mounted.

Additionally, the lens mount unit 11 may include a lens fastening unit 11b that secures the mounted lens 11a.

The lens fastening part 11b may be any fastening means that can be fastened to the lens mount part 11.

Additionally, the lens mount portion 11 has a penetrating center, and a transparent plate 11c may be coupled to the penetrating portion to prevent foreign substances from entering the image sensor 13a.

At this time, the transparent plate 11c can be made of any material as long as it is made of various transparent materials such as glass and plastic.

Additionally, the lens mount unit 11 can be assembled with the inside of the sensor mechanism unit 17b, which will be described later. At this time, assembly is not only possible through fastening through threads, but it may also be possible to use separate bolts.

The lighting unit 12 can adjust wavelength or brightness. Additionally, the lighting unit 12 may be selectively used in various types depending on the type and quantity of lighting.

In addition, the lighting unit 12 may be coupled to one side and to the top and bottom in the direction of the lens mount unit 11, respectively.

In addition, the lighting unit 12 may include a lighting connector 12a on the other side that protrudes in the direction of the sensor unit 13 and is electrically coupled to the sensor unit 13.

The sensor unit 13 may receive light incident on the lens as it passes through the lens mount unit 11. This sensor unit 13 may include an image sensor (13a).

Additionally, the sensor unit 13 can be selectively used in various types depending on the resolution of the sensor.

Additionally, the sensor unit 13 may have a first sensor connector 13b protruding from one upper side so as to be electrically coupled to the lighting connector 12a.

Additionally, the sensor unit 13 may have a second sensor connector 13c protruding in the direction of the FPGA unit 14 on one side of the bottom to be electrically coupled to the FPGA unit 14.

Additionally, the image sensor 13a may be located between the first sensor connector 13b and the second sensor connector 13c. That is, the image sensor 13a can be positioned to correspond to the transparent plate 11c.

The FPGA unit 14 can verify image processing operation and performance. Additionally, the FPGA unit 14 can be selectively used in various types depending on performance and capacity.

Additionally, the FPGA unit 14 may have a first FPGA connector 14a protruding on one upper side in the direction of the embedded unit 15 so as to be electrically coupled to the embedded unit 15.

Additionally, the FPGA mechanism unit 17c may have a second FPGA connector 14b protruding on one side of the lower end in the direction of the second sensor connector 13c to be electrically coupled to the second sensor connector 13c.

The embedded unit 15 can analyze the image and receive the results. Additionally, the embedded portion 15 can be selectively used in various types depending on performance and function.

Additionally, the embedded portion 15 may have a first embedded connector 15a protruding on one upper side in the direction of the first FPGA connector 14a so as to be electrically coupled to the first FPGA connector 14a.

Additionally, the embedded portion 15 may have a second embedded connector 15b formed on one bottom side protruding in the direction of the interface portion 16 to be electrically coupled to the interface portion 16.

The interface unit 16 can send an output signal of the result value analyzed by the embedded unit 15 to the outside. Additionally, the interface unit 16 can be selectively used in various types depending on the output method, such as USB or GigE.

Additionally, the interface unit 16 may have an interface connector 16a formed on one lower side thereof to be electrically coupled to the second embedded connector 15b.

Additionally, the interface portion 16 may include an interface coupling portion 16b on one upper side that protrudes in the direction of the rear cover to be coupled to the rear cover.

Meanwhile, the sensor unit 13, the FPGA unit 14, the embedded unit 15, and the interface unit 16 may be composed of a PCB BOARD.

That is, the mechanism part 17, which will be described later, can be fastened to the top and bottom of the PCB BOARD, respectively.

In addition, a protective cap (not shown) may be formed inside the mechanism 17 to prevent damage to the PCB BOARD at a place where the mechanism 17 and the PCB BOARD come into contact.

These protective caps are made of elastic materials and can protect the PCB BOARD from external shocks.

Additionally, the machine vision camera 1 based on deep learning may further include a rear cover portion 20 formed according to the shape of the interface portion 16.

In addition, the module unit 10 may include a mechanical unit 17 configured to enable assembly.

The mechanism portion 17 may include a lighting mechanism portion 17a, a sensor mechanism portion 17b, an FPGA mechanism portion 17c, an embedded mechanism portion 17d, an interface mechanism portion 17e, and a back cover mechanism portion 18f. there is.

In addition, the lighting device unit 17a may include lighting uneven parts 17a-1 on one of the upper and lower sides, respectively, that protrude in the direction of the sensor unit 13 to be coupled to the sensor unit 13.

The sensor mechanism portion 17b may be coupled to the upper and lower ends respectively to fix the sensor portion 13.

In addition, the sensor mechanism portion 17b may be formed with a first sensor groove 17b-1 coupled to the lighting uneven portion 17a-1 on one side of the upper and lower sides, respectively.

Additionally, the sensor mechanism portion 17b may have a second sensor groove 17b-2 formed in the direction of the FPGA mechanism portion 17c, which is opposite to the first sensor groove 17b-1 formed on one side of the upper and lower sides.

The FPGA mechanism 17c may be coupled to the upper and lower ends respectively to secure the FPGA mechanism 17c at both ends.

In addition, the FPGA mechanism unit 17c may be formed with an FPGA convex-convex portion 17c-1 on one side of the upper and lower sides, respectively, which protrudes to be coupled to the first sensor concave-convex groove 17b-1.

Additionally, the FPGA mechanism portion 17c may have an FPGA convex-convex groove 17c-2 formed in the direction of the embedded mechanism portion 17d, which is opposite to the FPGA convex-convex portion 17c-1 formed on one side of the upper and lower sides.

The embedded mechanism portion 17d may be coupled to the upper and lower ends respectively to secure the embedded portion 15 at both ends.

Additionally, the embedded mechanism portion 17d may include an embedded concave-convex portion 17d-1 that protrudes to be coupled to the FPGA concave-convex groove 17c-2 on one side of the upper and lower sides, respectively.

Additionally, the embedded mechanism portion 17d may have an embedded concave-convex groove 17d-2 formed in the direction of the interface mechanism portion 17e, which is opposite to the embedded concavo-convex portion 17d-1 formed on one side of the upper and lower sides.

The interface mechanism portion 17e may be coupled to the upper and lower ends respectively to secure the interface portion 16 at both ends.

In addition, the interface mechanism portion 17e may include an interface concave-convex portion 17e-1 that protrudes to engage the embedded concave-convex groove 17d-2 on one side of the upper and lower sides, respectively.

In addition, the interface mechanism portion 17e may include an interface convex-convex groove 17e-2 on one side of the upper and lower sides, respectively, in the direction of the rear cover opposite to the interface concave-convex portion 17e-1.

The rear cover portion 20 may include a rear cover uneven portion 18f-1 that protrudes in the direction of the interface uneven groove 17e-2 so as to be coupled to the interface uneven groove 17e-2.

Meanwhile, the interior of the mechanism unit 17 located at the upper and lower ends of the mechanism unit 17 may be formed through penetration to enable bolt connection.

At this time, the bolt can be divided into a bolt for coupling lighting (19a) and a bolt for coupling a camera (19b).

Here, the lighting coupling bolt (19a) penetrates the lighting fixture portion (17a) and the sensor mechanism portion (17b), but a portion of the sensor mechanism portion (17b) penetrates, and the camera coupling bolt (19b) extends from the rear cover portion (20). It may penetrate up to a portion of the sensor unit 13.

However, this is only an example, and the mechanism parts can be coupled by passing all of them through the lighting coupling bolt (19a) or the camera coupling bolt (19b).

Figures 7 and 8 are diagrams showing a machine vision camera based on deep learning according to another embodiment of the present invention.

Referring to FIGS. 7 and 8 , it is composed of individual modules and may include a module unit 100 for selecting and assembling each module, and a coupling unit 200 for combining the module units 100.

At this time, the module unit 100 includes a lens 11a assembled, a lens mount unit 11 that can be assembled according to the specifications of the lens 11a, a lighting unit 12 that adjusts the wavelength or brightness, and a lens 11a. a sensor unit 13 that receives light passing through the lens mount unit 11, an FPGA unit 14 that verifies image processing operation and performance, and an embedded unit that analyzes the image and receives the results ( 15) and an interface unit 16 that sends an output signal of the result value analyzed by the embedded unit 15 to the outside.

Additionally, the coupling portion 200 may include a rear cover portion 20 coupled to the sensor portion 13.

Additionally, the coupling unit 200 can combine the lens mount unit 11, the lighting unit 12, the sensor unit 13, the FPGA unit 14, the embedded unit 15, and the interface unit 16. there is.

This allows the coupled state to be fixed through the bolt portion.

At this time, the detailed coupling and electrical coupling of the lens mount unit 11, the lighting unit 12, the sensor unit 13, the FPGA unit 14, the embedded unit 15, and the interface unit 16 are It may be the same as the above-described technology described with reference to FIGS. 1 to 6.

Therefore, through the machine vision camera 10 based on deep learning according to the present invention, the optimal smart camera is created by combining each module in a modular manner to suit the usage environment. It can be produced.

In other words, more than hundreds of combinations can be created, providing smart cameras suitable for various manufacturing environments.

In addition, even if assembly has already been completed and installed on site, it can be easily replaced when one or multiple modules need to be changed.

This allows manufacturers to create customized smart cameras that meet the diverse needs of customers.

Optimal embodiments are disclosed in the drawings and specification. Here, specific terms are used, but they are used only for the purpose of explaining the present invention and are not used to limit the meaning or scope of the present invention described in the claims. Therefore, those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Therefore, the true technical protection scope of the present invention should be determined by the technical spirit of the attached patent claims.

1: Machine vision camera based on deep learning
10: module part
11: Lens mount part
11a: lens
11b: Lens fastening part
11c: Transparent plate
12: lighting unit
12a: Lighting connector
13: sensor unit
13a: Image sensor
13b: first sensor connector
13c: 2nd sensor connector
14: FPGA part
14a: 1st FPGA connector
14b: 2nd FPGA connector
15: Embedded part
15a: 1st embedded connector
15b: 2nd embedded connector
16: interface part
16a: Interface connector
16b: Interface coupling part
17: Mechanism part
17a: Lighting device part
17a-1: Lighting uneven part
17b: Sensor mechanism part
17b-1: First sensor concavo-convex groove
17b-2: Second sensor uneven groove
17c: FPGA mechanism part
17c-1: FPGA concavo-convex portion
17c-2: FPGA uneven groove
17d: embedded mechanism
17d-1: Embedded uneven part
17d-2: Embedded uneven groove
17e: Interface mechanism part
17e-1: Interface uneven portion
17e-2: Interface uneven groove
18f: Rear cover mechanism part
18f-1: Rear cover uneven part
19a: Bolt for combining lighting
19b: Camera coupling bolt
20: Rear cover part
100: module part
200: coupling part

Claims (1)

It is composed of each module and includes a module part for selecting and assembling each module,
The module part,
A lens mount unit on which a lens is assembled and can be assembled according to the specifications of the lens,
A lighting unit that controls wavelength or brightness,
a sensor unit that receives light incident on the lens as it passes through the lens mount unit and includes a Youngsan sensor;
An FPGA unit that verifies image processing operation and performance,
An embedded unit that analyzes the video and receives the results,
an interface unit that sends an output signal from the result analyzed by the embedded unit to the outside;
A mechanism configured to enable assembly and
It includes a rear cover portion formed according to the shape of the interface,
The lens mount portion includes a mounting portion on which the lens is mounted, a lens fastening portion that secures the lens mounted on the mounting portion, and a central portion formed through the penetrating portion to prevent foreign substances from entering the image sensor. It includes a transparent plate coupled to the portion,
The lighting unit includes a lighting connector on the other side that protrudes in the direction of the sensor unit and is electrically coupled to the sensor unit,
The sensor unit has a first sensor connector protruding on one upper side to be electrically coupled to the lighting connector, and a second sensor connector protruding in the direction of the FPGA unit on one lower side to be electrically coupled to the FPGA unit,
The image sensor is located between the first sensor connector and the second sensor connector and is positioned to correspond to the transparent plate,
A first FPGA connector is formed on one upper side of the FPGA unit to protrude in the direction of the embedded part to be electrically connected to the embedded part,
The FPGA mechanism unit has a second FPGA connector protruding on one side of the lower end in the direction of the second sensor connector to be electrically coupled to the second sensor connector,
The embedded part has a first embedded connector protruding in the direction of the first FPGA connector to be electrically coupled to the first FPGA connector on one side of the upper part, and a second embedded connector in the direction of the interface part to be electrically coupled to the interface part on one side of the lower part. A protrusion is formed,
The interface unit includes an interface connector formed on one side of the lower end to be electrically coupled to the second embedded connector, and an interface coupling part formed on one side of the upper end that protrudes in the direction of the rear cover to be coupled to the rear cover,
The sensor unit, FPGA unit, embedded unit, and interface unit are composed of a PCB BOARD,
The mechanism part is fastened to the top and bottom of the PCB BOARD, and a protective cap is formed inside the mechanism part to prevent damage to the PCB BOARD at the place where the PCB BOARD is fastened and comes into contact,
The mechanism part includes a lighting mechanism part, a sensor mechanism part, an FPGA mechanism part, an embedded mechanism part, an interface mechanism part, and a back cover mechanism part,
The lighting fixture unit includes lighting irregularities on one side of the upper and lower sides, respectively, that protrude in the direction of the sensor unit to be coupled to the sensor unit,
The sensor mechanism unit is coupled to the upper and lower ends to fix the sensor unit, and a first sensor concave-convex groove coupled to the lighting concave-convex portion is formed on one side of the upper and lower portions, and a first sensor concavo-convex groove is formed on one side of the upper and lower sides. A second sensor groove is formed in the direction opposite to the FPGA mechanism,
The FPGA mechanism is coupled to the top and bottom to secure the FPGA mechanism to both ends, and an FPGA concave-convex portion is formed to protrude to be coupled to the first sensor concave-convex groove on one side of the upper and lower sides, respectively, and an FPGA convex-convex portion is formed on one side of the top and bottom. An FPGA concave-convex groove is formed in the direction of the embedded mechanism, which is opposite to the formed FPGA concavo-convex portion.
The embedded mechanism may be coupled to the upper and lower ends respectively to fix the embedded parts at both ends, and includes embedded concavo-convex portions protruding and formed on one side of the upper and lower sides to engage with the FPGA concavo-convex grooves, and embedded portions formed on one side of the upper and lower sides. An embedded concavo-convex groove is formed in the direction of the interface mechanism, which is opposite to the concavo-convex portion,
The interface mechanism part is coupled to the upper and lower ends to fix the interface parts at both ends, respectively, and includes an interface concavo-convex portion protruding to engage with an embedded concave-convex groove on one side of the upper and lower sides, respectively, and an interface concavo-convex portion on one side of the upper and lower portions, respectively. It includes an interface concave-convex groove in the opposite direction of the rear cover,
The rear cover portion includes a rear cover concavo-convex portion that protrudes in the direction of the interface concave-convex groove so as to be coupled to the interface concavo-convex groove,
The interior of the mechanism located at the top and bottom of the mechanism is formed through a hole to enable bolt connection, and is divided into bolts for combining lighting and bolts for combining cameras,
The lighting coupling bolt penetrates the lighting fixture portion and the sensor mechanism portion, but passes through a portion of the sensor mechanism portion, and the camera coupling bolt penetrates from the rear cover portion to a portion of the sensor portion. camera.