KR20210106210A - Visually Impaired Gait Assistance Based on AI Learning System and Method - Google Patents

Abstract

A visually impaired walking assistance system according to the present invention may comprise: a neckband including front cameras for far/near distance and left/right cameras, to take pictures of surrounding situations while the visually impaired are walking, a speaker for providing guidance information, LED for serving as a warning, guidance light or the like, etc.; a smartphone including a walking assistance application for the visually impaired that calculates objects, signals, characters, and distance measurement information necessary for walking of the visually impaired based on the model data learned through artificial intelligence learning by using the image captured by the camera of the neckband, and combines the frequency of occurrence and risk level of each event based on the objects, signals, characters, and distance measurement information to calculate the weight of the risk object and use the weight information as a priority selection criterion for user warning voice guidance between walking; and a remote support center including real-time remote support function related to walking at a user's request.

Description

Visually Impaired Gait Assistance Based on AI Learning System and Method

The present invention provides a walking assistance system for the visually impaired consisting of a neckband, a smartphone app, and a remote support center, and an image captured by a camera in the system for the walking of the visually impaired based on model data learned through artificial intelligence learning. Blind person who calculates necessary object, signal, text and distance measurement information, calculates risk object weight by combining the frequency of occurrence and risk of each event, and uses the weight information as a priority selection criterion for user warning voice guidance between walking It relates to a method of walking assistance.

It is difficult for blind people with limited visual function to freely engage in normal activities in daily life or walking. In addition, in our society, there are great restrictions on the activities of the visually impaired due to the poor development of assistive devices along with poor social prejudice and poor welfare system related to the visually impaired.

Accordingly, there are Braille blocks, voice guidance devices, and sticks as protection devices for the visually impaired that allow the visually impaired to safely walk without inconvenience in places and facilities where they can perform normal activities.

In addition, the conventional cane is a protection device for the visually impaired that is mainly used by the visually impaired while walking, and it is difficult to detect floating obstacles on the ground as well as providing current location information and effective protection measures in case of an emergency or emergency to the visually impaired walking. The problem is that it cannot be done.

Recently, guide dogs have been trained so that they can walk safely with the visually impaired, but these measures are too realistic for the visually impaired to use easily, given the cost investment for training guide dogs and additional facilities to maintain and protect guide dogs. No problem occurred.

Accordingly, there is a need for a method for protecting the visually impaired that can protect the visually impaired as well as efficiently protect the visually impaired in case of an emergency or emergency.

Accordingly, the technical problem of the present invention was conceived in this regard, and an object of the present invention is to provide a camera for photographing the surrounding situation when a visually impaired person walks, a speaker for providing guidance information, and a warning or guidance light. a neckband comprising an LED light; Using the image captured by the camera of the neckband, based on the model data learned through artificial intelligence learning, objects, signals, characters, and distance measurement information necessary for walking of the visually impaired are calculated, and the recognized objects, signals, characters , a walking support app for the visually impaired that guides walking by combining the frequency of occurrence and risk of each event based on distance measurement to calculate the risk object weight and using the weight information as a priority selection criterion for user warning voice guidance between walking including a smartphone; and a remote assistance center including a walking-related real-time remote assistance function at the user's request.

The present invention provides a method of calculating objects, signals, texts and distance measurement information necessary for walking of the visually impaired based on model data learned through artificial intelligence learning using images captured in a walking assistance system for the visually impaired and guiding them; Based on the recognized object, signal, text, and distance measurement, the risk object weight is calculated by combining the occurrence frequency and risk level of each case, and the weight information is used as a priority selection criterion for user warning voice guidance between walking to guide walking. it's about how to

The blind walking assistance system according to the present invention provides a far/near front camera, a left/right camera, a speaker for providing guidance information, and an LED that serves as a warning or guide light for photographing surrounding situations when a visually impaired person walks. A neckband comprising: Calculating objects, signals, texts, and distance measurement information necessary for walking of the visually impaired based on model data learned through artificial intelligence learning using the image captured by the camera of the neckband, and Based on the recognized object, signal, text, and distance measurement, the frequency of occurrence of each event and the degree of risk are combined to calculate the risk object weight, and the weight information is used as a priority selection criterion for user warning voice guidance between walks to guide walking A smart phone including a walking assistance app for the visually impaired; and a remote support center including a walking-related real-time remote support function at the user's request.

In one embodiment, the walking support app of the smartphone is a text-to-speech app that converts text recognized as a request of a pedestrian into voice and a remote support worker responds to the request of a pedestrian and enables real-time walking-related support. It may include a remote support app.

In one embodiment, the remote support center includes a remote support program, the user's location and image are shared with the remote support personnel at the user's request, and the remote support personnel responds to the pedestrian's request to provide real-time assistance for walking. It may be possible.

In one embodiment, the walking support app of the smartphone may perform a distance calculation algorithm according to the pixel position with the image captured by the front short-distance camera of the neckband.

In one embodiment, the walking support app of the smartphone may perform a direction calculation algorithm according to pixel positions with images captured by the front camera for far/near distance and the left/right camera of the neckband.

In one embodiment, In the walking support app of the smartphone, the distance calculation method by the distance calculation algorithm to the object after object identification may be “distance to object = distance on image-processing delay * movement speed”.

In an embodiment, an image integration process for merging images captured by the front camera and the left/right camera for the far/near distance of the neckband may be performed.

In one embodiment, in the walking support app of the smartphone, the image integration ratio for merging images taken from the front camera, the left camera and the right camera of the neckband is adjusted and transmitted to the image receiver of the smartphone. and the adjusted image integration ratio can be applied to the image integration processing process of the neckband.

In one embodiment, the image integration ratio according to the dangerous object recognition is 1:1:1:1 in a general situation for the front (near), front (far), left, and right cameras, the front near-danger object approach It can be adjusted to 3:1:1:1 in case of situation, 1:1:2:2 in case of approaching rear/side dangerous object, and 1:3:1:1 in case of approaching remote dangerous object in front. .

The blind walking assistance method according to the present invention comprises the steps of: when the blind person starts walking, taking an image from a front camera for far/near and a left/right camera of a neckband worn on the neck of the visually impaired; integrating the four captured images into a frame form; Receiving the image in the form of an integrated frame through wireless communication to a walking support app of a smart phone; Using the received image, the object, signal, text and distance measurement information necessary for walking of the visually impaired is calculated based on the model data learned through artificial intelligence learning in the walking support app of the smartphone, and the recognized object; Based on the signal, text, and distance measurement, the frequency of occurrence of each event and the degree of risk are combined to calculate the risk object weight, and the weight information is used as a priority selection criterion for user warning voice guidance to generate voice guidance information to be provided to pedestrians. to do; transmitting the generated voice guide information to a controller of a neckband; and operating a speaker of the neckband by a controller of the neckband based on the received voice guidance information to transmit the voice guidance information to the pedestrian by voice.

In one embodiment, the walking support app of the smartphone may perform a distance calculation algorithm according to the pixel position with the image captured by the front short-distance camera of the neckband.

In one embodiment, the walking support app of the smartphone may perform a direction calculation algorithm according to pixel positions with images captured by the front camera for far/near distance and the left/right camera of the neckband.

In an embodiment, in the walking support app of the smartphone, the distance calculation method by the distance calculation algorithm to the object after identifying the object may be “distance to object = distance on image-processing delay * movement speed”.

In an embodiment, an image integration process for merging images captured by the front camera and the left/right camera for the far/near distance of the neckband may be performed.

In one embodiment, the walking support app of the smartphone adjusts the image integration ratio for merging images captured by the front camera for far/near distance and the left/right camera of the neckband, and transmits it to the image receiver of the smartphone. The adjusted image integration ratio may be applied to the image integration processing process of the neckband.

In one embodiment, the image integration ratio according to the dangerous object recognition is 1:1:1:1 in the case of a general situation for the far/near front camera and the left/right camera, in the case of a front near-danger object approach situation can be adjusted to 3:1:1:1, 1:1:2:2 in case of rear/side dangerous object approach situation, and 1:3:1:1 in front remote dangerous object approach situation.

According to embodiments of the present invention, the blind walking assistance system can calculate objects, signals, texts, and distance measurement information necessary for walking of the visually impaired and effectively guide them for walking of the visually impaired.

In addition, the risk object weight is calculated by combining the occurrence frequency and risk level of each event based on the recognized object, signal, text, and distance measurement, and the weight information is used as a priority selection criterion for user warning voice guidance between walking. Information with high risk priority can be given priority to the visually impaired, so the ability to cope with risks can be greatly improved.

In addition, it is possible to provide real-time support linked to a remote support center based on the location and surrounding image information of the weak walking person, thereby providing a guide system for the visually impaired with improved safety than the risk.

1 is a schematic explanatory diagram for explaining a system for assisting the visually impaired and its main functions according to an embodiment of the present invention.
2 is a schematic design diagram of a neckband device among the walking assistance system for the visually impaired according to an embodiment of the present invention.
3 is a block diagram of main equipment of a system for assisting the visually impaired according to an embodiment of the present invention.
4 shows a criterion for selecting risk object weight information by combining the occurrence frequency and risk level based on the user's gait information by the artificial intelligence learning model in the walking assistance system for the visually impaired according to an embodiment of the present invention.
5 is a flowchart of a gait support function processing in a system for assisting the visually impaired according to an embodiment of the present invention.
6 illustrates a series of processes for an object recognition method using an artificial intelligence learning model in a walking assistance system for the visually impaired according to an embodiment of the present invention.
7 illustrates an example of object recognition results by an artificial intelligence learning model in the walking assistance system for the visually impaired according to an embodiment of the present invention.

Since the present invention may have various changes and may have various forms, embodiments will be described in detail in the text. However, this is not intended to limit the present invention to the specific disclosed form, it should be understood to include all modifications, equivalents and substitutes included in the spirit and scope of the present invention. In describing each figure, like reference numerals have been used for like elements. Terms such as first, second, etc. may be used to describe various elements, but the elements should not be limited by the terms.

In the present application, terms such as "comprises" or "consisting of" are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification is present, but one or more other features It is to be understood that it does not preclude the possibility of the presence or addition of numbers, steps, operations, components, parts, or combinations thereof.

Unless defined otherwise, all terms used herein, 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. Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the accompanying drawings.

1 is a schematic explanatory diagram for explaining a walking assistance system 1 for the visually impaired and main functions thereof according to an embodiment of the present invention.

1 is a walking assistance system 1 for the visually impaired according to an embodiment of the present invention by analyzing the image received from the neckband 100 including a camera and a speaker, and the neckband 100 through artificial intelligence learning. It may include a remote support center 300 including a smart phone 200 that provides information necessary for gait and a real-time remote support function related to walking by a user's request.

The main functions of the blind walking assistance system 1 are functions such as object recognition to guide the visually impaired walking, recognition of various traffic signals, recognition/measurement of distances between objects, and character recognition of surrounding objects.

The performance of the system may be determined depending on how accurately the functions and the like are implemented and helpful for the visually impaired.

The blind walking assistance system of the present invention analyzes images captured by a neckband camera based on model data learned through artificial intelligence learning to improve the performance of these functions and increase accuracy.

2 is a schematic design diagram of the neckband 100 device of the walking assistance system 1 for the visually impaired according to an embodiment of the present invention.

Referring to FIG. 2 , the neckband device 100 of the walking assistance system 1 for the visually impaired according to the present embodiment includes two front cameras 101 and 102 , two side cameras 103 and 104 , and four infrared rays. LEDs (105, 106, 107, 108), 2 speakers (109, 110), 2 LED indicators (111, 112), 2 LED modules (113, 114), image processing board 115, control board 116 , a power board 117 , and a battery 118 may be included.

In this case, the type, shape, structure and number of the camera, infrared LED, LED indicator, LED module, and speaker are not limited.

On the other hand, by selecting a device having a neckband-type structure as a walking assistance device for the visually impaired, the visually impaired person wears the neckband-type device around the neck and uses the four cameras installed on the neckband when walking. It is possible to take pictures freely without blocking or interference.

In addition, the visually impaired who wear it can always use their hands freely, so that more convenience can be pursued while walking.

The front cameras 101 and 102 of the neckband may be installed at one end of the neckband, one for a long distance and one for a short distance.

The images captured by the four cameras may be processed by the image processing board 115 .

Accordingly, infrared LEDs 105 and 106 are installed in accordance with each of the distant and short-distance cameras, so that the camera can be photographed even at night when there is no light.

In addition, the side cameras (103, 104) are installed one by one on both sides of the neckband 100 can be used for photographing the left and right parts of the visually impaired.

Accordingly, infrared LEDs 107 and 108 may be installed to match the left and right cameras.

The two speakers 109 and 110 are located one on each side of the neckband 100, but if they are located at a distance close to a person's ear, it may be effective in terms of transmitting sound.

The two LED indicators 111 and 112 are located one on each side of the neckband 100, but if they are located close to the human eye, it is effective to recognize the LED light even for the visually impaired.

The two LED indicators can be controlled by the LED modules 113 and 114 present in the neckband 100 .

The two LED indicators 111 and 112 may be used for warning or notification to the visually impaired.

The battery 118 may be controlled by the power board 117 and may provide power required for the neckband 100 .

A control unit 120 for controlling the image processing board 115 , the LED modules 113 and 114 , the speakers 109 and 110 , and the power board 117 is located on the control board 113 .

3 is a block diagram of the main equipment of the walking assistance system 1 for the visually impaired according to an embodiment of the present invention.

Referring to FIG. 3 , four cameras including two front cameras 101 and 102 and left/right cameras 103 and 104 of the neckband 100 are connected to the image processing board 115 and connected to the image processing board ( 115) can be controlled.

The images captured by the four cameras 101 , 102 , 103 , and 104 may be merged into an image integrated frame on the image processing board 115 , and a processing process such as adjusting the image integration ratio may be performed.

Accordingly, infrared LEDs 105, 106, 107, 108) may be installed to fit each of the four cameras 101, 102, 103, and 104, which may be operated under the control of the illuminance sensor 119 present in the image processing board 115.

The infrared LED has the advantage of enabling the camera to take pictures even at night when there is no light.

That is, infrared LEDs invisible to the human eye can be used as lighting by using infrared LEDs. Although the photographed image only appears in black and white, it has superior nighttime shooting ability compared to a camera with general lighting, so it can be effective when the visually impaired move at night or use a dark place.

As described above in FIG. 2 , in the neckband 100 , two speakers 109 and 110 and two LED indicators 111 and 112 are provided at predetermined positions of the neckband 100 .

The two LED indicators 111 and 112 can be controlled by the LED modules 113 and 114 present in the neckband.

In addition, the battery 118 may be controlled by the power board 117 to provide power required for the neckband 100 .

A control unit 120 for controlling the image processing board, the LED modules 113 and 114 , the speakers 109 and 110 , and the power board 117 is located on the control board 113 . In addition, the control board 116 may include a communication unit 121 for communicating with the outside of the neckband and a microphone 122 for acquiring a user's voice.

By driving the communication unit 121 of the neckband 100 , image information captured by the camera of the neckband 100 may be transmitted to the smartphone 200 .

The smartphone 200 includes a walking support app 210, and analyzes the image received from the neckband 100 based on the model data learned through artificial intelligence learning, and an object necessary for walking of the visually impaired. /Obstacle recognition, signal recognition, and character recognition functions can be performed.

In this case, a distance calculation algorithm according to an image pixel position and a direction calculation algorithm according to a pixel position may be used.

The smart phone 200 walking support app 210 may include a text-to-speech app 220 for converting the recognized text into speech.

The smart phone 200 walking support app 210 may include an SOS remote support app 230 that enables remote support personnel to provide real-time walking-related support in response to a pedestrian's request.

The control information generated by the smartphone walking support app 210 is transmitted to the communication unit 121 of the neckband 100 through the communication unit (not described here) of the smartphone 200, and the neckband receiving it The controller 120 of 100 may be used to control each associated device.

That is, for example, if the text for the subway station entrance is recognized by the text-to-speech app 220 of the smartphone 200 and voice guidance information is prepared for the user, this information is transmitted to the controller 120 of the neckband 100 and the control unit 120 may be controlled such that the information is guided through the speakers 109 and 110 of the neckband 100 .

The remote support center 300 includes a remote support program 310 to share the location and image of the user with the remote support personnel at the request of the user, and the remote support personnel responds to the pedestrian's request to provide real-time assistance related to walking. possible.

4 shows a criterion for selecting risk object weight information by combining the occurrence frequency and risk level based on the user's gait information by the artificial intelligence learning model in the walking assistance system 1 for the visually impaired according to an embodiment of the present invention. will be.

This is to calculate the risk object weight by combining the occurrence frequency and risk level based on the user's walking information using artificial intelligence, and the risk object weight information is used as a priority selection criterion for user information between walking.

Referring to FIG. 4 ( 1 ), in one embodiment, with respect to the criterion for the frequency of occurrence, if the occurrence of an event occurs more than once a month by analyzing user gait information, the occurrence frequency is set as low, and if it occurs more than once a week The frequency is set as moderate, and if it occurs more than once a week, it can be set as a high frequency of occurrence.

In addition, referring to FIG. 5 ( 2 ), in one embodiment, with respect to the standard of the level of risk, if it is a risk other than a physical risk of a vehicle parked on the roadway, the risk level is set as low, and oncoming bicycles and walking dents If it is a risk that can cause a minor bodily hazard, such as a location, the risk level is considered as moderate, and if it is a risk that can cause serious bodily harm, such as a construction site or oncoming vehicle on a sidewalk, the risk level may be set as high.

Based on the criteria for the occurrence frequency and risk, as shown in FIG. 5(3), the weight of the walking risk object may be calculated.

That is, the weight of each walking risk object can be calculated by setting the frequency of occurrence as low, normal, and high and the risk as low, normal, and high. By using the set weight of the gait risk object, it is possible to determine the priority of the alarm, that is, the voice guidance when the visually impaired walk.

That is, the higher the weight of the walking risk object, the higher the priority of the voice guidance, and accordingly, the voice guidance can be performed before other voice guidance.

If the weight of the walking risk object is low, the voice guidance is determined as a lower priority in priority, and accordingly, the voice guidance can be performed with a lower priority than other voice guidance.

5 is a view showing a walking support in the walking assistance system 1 for the visually impaired while calculating the risk object weight by combining the occurrence frequency and risk level of each event of FIG. 4 and using the weight information as a priority selection criterion for user warning voice guidance A flowchart of the functional processing is shown.

When the blind person starts walking, the four cameras 101, 102, 103, 104 of the neckband worn on the neck of the visually impaired take images in the near, far, forward, left and right directions (S1).

The four captured images are integrated in the form of a frame (S2).

The image in the form of the integrated frame is received by the image receiving unit of the smart phone 200 through wireless communication (S3).

The image analysis unit existing in the walking support app of the smartphone 200 analyzes the image received from the neckband 100 based on the model data learned through artificial intelligence learning (S5).

Here, it is possible to perform a distance calculation algorithm according to the identification of the dangerous object and the pixel position with the image taken by the front camera 101 for a short distance.

As the distance calculation algorithm for the object, the method of calculating the distance to the object may be “distance to object = distance on image-processing delay * movement speed” (S5-1).

As an example, by calculating the risk object identification and distance, calculating the risk object weight by combining the occurrence frequency and risk of each event, and using the weight information as a priority selection criterion for user warning voice guidance, the primary voice guidance target is selected If so, you can proceed with the first voice guidance. For example, a voice guidance such as "There are stairs 3 m ahead" can be performed.

After that, the direction calculation is carried out for each object with the images taken from the front remote camera and the left/right camera, and the list up is performed, and the risk object is identified and distance calculated. By calculating and using the weight information as a priority selection criterion for user warning voice guidance, a secondary voice guidance target may be selected and the secondary guidance may be performed. For example, a voice guidance with direction information may be provided saying, “There is a crosswalk at 2 o'clock ahead.”

In addition, it is possible to determine the situation in which the dangerous object price approaches by calculating the size and position change for each dangerous object by comparing with the previous frames of the images captured by the front remote camera and the left/right camera.

For this reason, the dangerous object identification and distance are calculated, and the risk object weight is calculated by combining the occurrence frequency and risk of each event. A bicycle is approaching in the o’clock direction.”

In addition, the walking support app of the smartphone may perform a direction calculation algorithm according to the pixel position with the images captured by the front and left cameras of the neckband and the left/right cameras (S5-2).

The voice guidance information is transmitted from the smart phone to the controller of the neckband, so that the voice guidance information can be received from the neckband (S6-S7).

The speakers 109 and 110 of the neckband may receive the voice guidance information from the smartphone 200 and automatically adjust the volume based on the measurement of ambient noise to provide voice guidance (S8).

Using the received voice guidance information, the control unit of the neckband operates the speaker 1 and the speaker 2 so that the guidance message can be delivered to the user by voice (S9).

As illustrated, in the image analysis unit present in the walking support app of the smartphone 200, the distance and direction to the dangerous object based on the image received from the neckband 100 based on the model data learned through artificial intelligence learning. It can help the visually impaired walk more accurately and safely by detecting changes in their position.

In addition, after the image analysis (S5), the image integration ratio of merging images captured by the front camera, the left camera, and the right camera of the neckband for far and near can be adjusted using a model through artificial intelligence learning. (S10).

The video integration ratio is 1:1:1:1 for the front (near), front (far), left, and right cameras in general situations, and 3:1:1:1 in the case of front near-danger object approach situations. , it can be adjusted to 1:1:2:2 in case of approaching rear/side dangerous object, and 1:3:1:1 in case of approaching remote dangerous object in front.

The adjusted image integration ratio may be transmitted from the smartphone to the image processing unit of the neckband and reflected in the image integration process (S11-S12).

6 illustrates a series of processes for an object recognition method using an artificial intelligence learning model in the walking assistance system for the visually impaired according to an embodiment of the present invention.

FIG. 6 is a method for calculating a risk object weight by combining the frequency of occurrence and risk of each event of FIG. 4, and using the weight information as a priority selection criterion for user warning voice guidance, as shown in FIG. 5, the blind walking assistance system (1) ) may correspond to a series of examples of the result of recognition and voice guidance through a flow of gait support function processing for a series of gait support.

First, referring to FIG. 6 ( 1 ), a lane and a vehicle in front can be recognized in the walking support app 210 of the smartphone using an artificial intelligence learning model through an image captured by the neckband 100 . At this time, the vehicle may be recognized as an obstacle in the forward direction of the visually impaired, so that voice guidance may be performed.

Referring to FIG. 6(2) , after a user passes a vehicle parked in a non-moving lane, a new obstacle in the distance outside the lane may be identified.

In addition, as shown in FIG. 6( 3 ), if there is a change in position, that is, a movement in the new obstacle, and it moves away from the walking direction of the visually impaired, the voice guidance may not proceed.

In addition, as shown in FIG. 6( 4 ), if there is a change in position, that is, a movement in the new obstacle, and it is recognized that approaching in the walking direction of the blind person is recognized, voice guidance for this may be performed.

7 illustrates an example of object recognition results by an artificial intelligence learning model in the walking assistance system for the visually impaired according to an embodiment of the present invention.

As illustrated in FIG. 7( 1 ) , distant and near objects may be recognized. That is, cars, trucks, people, and bicycles at a distance may be recognized, and a person and a handbag held by a person at a short distance may be recognized.

In addition, as shown in FIG. 7(2), distant cars, trucks, and traffic signals may be recognized, and vehicles, traffic signals and crosswalks in the near distance may be recognized.

Meanwhile, the user's location and image may be shared with remote support personnel at the request of the user through the remote support program of the remote support center.

In this case, the SOS remote support app of the smartphone may perform a real-time support function linked to the remote support center based on the user's location & surrounding image information.

That is, the remote support personnel can provide real-time assistance related to walking through the SOS remote support app of the smartphone by responding to the user's request based on the user's location & surrounding image information.

Although the above has been described with reference to the preferred embodiments of the present invention, those skilled in the art can variously modify and change the present invention without departing from the spirit and scope of the present invention as set forth in the claims below. You will understand that you can.

Neckband: 100, Smartphone: 200,
Remote Support Center: 300,
Walking assistance app: 210, Text-to-speech app: 220
SOS Remote Assistance App: 230

Claims (16)

A neckband including a front camera for far/near and a left/right camera, a speaker for providing guidance information, and an LED that serves as a warning or guide light for photographing surrounding situations when the visually impaired walk;
Using the image taken with the camera of the neckband, based on the model data learned through artificial intelligence learning, objects, signals, characters, and distance measurement information necessary for walking of the visually impaired are calculated, and the recognized objects, signals, A walking support app for the visually impaired that guides walking by combining the frequency of occurrence and risk of each event based on text and distance measurement to calculate the weight of the risk object and using the weight information as a priority selection criterion for user warning voice guidance between walking A smartphone comprising; and
A walking assistance system for the visually impaired including a remote support center including a real-time remote support function related to walking at the user's request. The SOS according to claim 1, wherein the walking support app of the smartphone enables a text-to-speech app that converts text recognized as a request of a pedestrian into voice and a remote support manpower responds to the request of a pedestrian to provide real-time support for walking. Blind walking assistance system comprising a remote assistance app. According to claim 1, The remote support center, including a remote support program, by the user's request to share the location and image of the user to the remote support personnel, and the remote support personnel to respond to the request of the pedestrian, walking-related real-time support is provided A blind walking assistance system characterized in that it is capable. The system of claim 1 , wherein the walking assistance app of the smart phone is capable of performing a distance calculation algorithm according to the pixel position with an image captured by a front short-distance camera of the neckband. According to claim 1, wherein the walking support app of the smart phone walking the visually impaired, characterized in that it performs a direction calculation algorithm according to the pixel position with images taken from the front camera for far/near distance and the left/right camera of the neckband. auxiliary system. The method of claim 1, wherein the distance calculation method by the distance calculation algorithm to the object after identifying the object in the walking support app of the smartphone is “distance to object = distance on image-processing delay * movement speed” blind walking assistance system. [2] The system of claim 1, wherein an image integration process for merging images captured by the front camera and the left/right camera of the neckband is performed. The method of claim 1, wherein the walking support app of the smartphone adjusts the image integration ratio for merging images captured by the front camera, the left camera, and the right camera of the neckband for far and near, and transmits it to the image receiver of the smartphone. and applying the adjusted image integration ratio to the image integration processing process of the neckband. The method according to claim 7, wherein the image integration ratio according to the dangerous object recognition is 1:1:1:1 in a general situation for the front (near), front (far), left, and right cameras, the front near-danger object approach Adjusted to 3:1:1:1 in case of situation, 1:1:2:2 in case of approaching rear/side dangerous object, and 1:3:1:1 in case of approaching distant dangerous object in front. A walking assistance system for the visually impaired. When the blind person starts walking, taking an image from a front camera for far/near and a left/right camera of a neckband worn on the neck of the visually impaired;
integrating the four captured images into a frame form;
Receiving the image in the form of an integrated frame through wireless communication to a walking support app of a smart phone;
Using the received image, the object, signal, text and distance measurement information necessary for walking of the visually impaired is calculated based on the model data learned through artificial intelligence learning in the walking support app of the smartphone, and the recognized object; Based on the signal, text, and distance measurement, the frequency of occurrence of each event and the degree of risk are combined to calculate the risk object weight, and the weight information is used as a priority selection criterion for user warning voice guidance to generate voice guidance information to be provided to pedestrians. to do;
transmitting the generated voice guide information to a controller of a neckband; and
and transmitting the received voice guidance information to a pedestrian by operating a speaker of the neckband by a controller of the neckband to transmit the voice guidance information to the pedestrian. 11. The method of claim 10, wherein the walking assistance app of the smart phone is capable of performing a distance calculation algorithm according to pixel positions with images captured by a front short-distance camera of the neckband. 11. The gait of claim 10, wherein the walking support app of the smart phone performs a direction calculation algorithm according to pixel positions with images captured by the front camera for far/near distance and the left/right camera of the neckband. auxiliary method. The method of claim 10, wherein the distance calculation method by the distance calculation algorithm to the object after identifying the object in the walking support app of the smartphone is “distance to object = distance on image-processing delay * movement speed” A method of assisting the visually impaired. 11. The method of claim 10, wherein an image integration process for merging images captured by the front camera and the left/right camera of the neckband for far/short distance is performed. The method according to claim 10, wherein the walking support app of the smartphone adjusts the image integration ratio for merging images captured by the front camera and the left/right camera of the neckband for far/short distance and transmits it to the image receiver of the smartphone. A method for assisting the visually impaired walk, characterized in that the adjusted image integration ratio is applied to the image integration processing process of the neckband. The method according to claim 10, wherein the image integration ratio according to the dangerous object recognition is 1:1:1:1 in a general situation for the far/near front camera and the left/right camera, and in the case of a front near-danger object approach situation. is 3:1:1:1, 1:1:2:2 in case of rear/side dangerous object approach situation, 1:3:1:1 in case of front remote dangerous object approach situation, characterized in that Blind walking assistance method.

Images