CN113689633A

CN113689633A - Scenic spot human-computer interaction method, device and system

Info

Publication number: CN113689633A
Application number: CN202110986296.9A
Authority: CN
Inventors: 郑勇; 陈海江
Original assignee: Zhejiang Lishi Technology Co Ltd
Current assignee: Zhejiang Lishi Technology Co Ltd
Priority date: 2021-08-26
Filing date: 2021-08-26
Publication date: 2021-11-23
Anticipated expiration: 2041-08-26
Also published as: CN113689633B

Abstract

The invention discloses a scenic spot human-computer interaction method, a device and a system, wherein the method comprises the following steps: pre-establishing a knowledge base aiming at a target scenic spot, wherein the knowledge base comprises: one or a combination of the playing knowledge, the traffic knowledge, the food culture knowledge and the popular characteristic knowledge in the scenic spot; the method comprises the steps of obtaining a voice ticket-taking instruction sent by a tourist, and analyzing the voice ticket-taking instruction of the tourist by using an NLP (line segment processing) module, wherein the voice ticket-taking instruction comprises the following steps: a ticket fetching instruction; carrying out authentication processing on the tourist and drawing a ticket; and acquiring a voice consultation instruction sent by a user, matching the voice consultation instruction based on a pre-established knowledge base, and outputting a matching result. The invention utilizes voice recognition to intelligently ask and answer, can also realize explanation on basic knowledge, basic environment, basic service and the like of scenic spots, and enhances the participation sense of tourists. The temperature is provided in the ticketing link, and the utilization rate of the self-service device is improved.

Description

Scenic spot human-computer interaction method, device and system

Technical Field

The invention relates to a man-machine interaction method, in particular to a scenic spot man-machine interaction method, a device and a system.

Background

With the development of tourism and the progress of modern science and technology, most scenic spots are provided with scenic spot automatic ticketing and marketing distribution systems, which play a role in relieving the pressure of scenic spot ticketing, reducing the time of tourists queuing for ticketing and improving the experience of tourists, wherein the self-service ticketing terminal mainly uses an automatic ticketing machine to realize the functions of inquiry, ticketing, payment and the like. The background system of the automatic ticket selling and taking machine comprises a main control module, a union pay card payment module and a ticket drawing module. The master control module is used for controlling the operation of the union pay card payment module and the ticket drawing module, the union pay card payment module is used for completing the fee deduction operation of the union pay card according to the fee deduction instruction of the master control module, and the ticket drawing module is used for completing the ticket drawing operation according to the ticket drawing instruction of the master control module.

However, there are two main problems in the prior art: 1. the self-service ticket selling and taking machine is not practical due to more carrying functions in the whole view, and has negative effects brought by complicated operation; 2. neglecting the emotional feeling of the tourists, having no interaction with the tourists, having low utilization rate, people still tend to manually ticket selling windows. And further, the input-output ratio of the self-service ticket selling and taking machine is not high, the investment is lost, and the comprehensive benefit of the scenic spot is low.

Disclosure of Invention

The invention aims to provide a scenic spot human-computer interaction method, device and system, and aims to solve the problem of low comprehensive benefit caused by low equipment utilization rate due to weak participation of tourists in the prior art.

The invention solves the technical problems through the following technical scheme:

the invention provides a scenic spot human-computer interaction method, which comprises the following steps:

pre-establishing a knowledge base aiming at a target scenic spot, wherein the knowledge base comprises: one or a combination of the playing knowledge, the traffic knowledge, the food culture knowledge and the popular characteristic knowledge in the scenic spot;

the method comprises the steps of obtaining a voice ticket-taking instruction sent by a tourist, and analyzing the voice ticket-taking instruction of the tourist by using an NLP (line segment processing) module, wherein the voice ticket-taking instruction comprises the following steps: a ticket fetching instruction; carrying out authentication processing on the tourist and drawing a ticket;

and acquiring a voice consultation instruction sent by a user, matching the voice consultation instruction based on a pre-established knowledge base, and outputting a matching result.

Optionally, the knowledge base of the target scenic spot further includes a customized knowledge base, and the establishing process of the customized knowledge base includes:

acquiring related knowledge, and classifying the related knowledge by using a clustering algorithm to obtain a plurality of knowledge classifications;

acquiring a scene label corresponding to each knowledge classification, and marking the corresponding knowledge classification by using the scene label, wherein the scene label comprises: meeting scene labels, research scene labels, hotel scene labels, natural knowledge scene labels, historical culture knowledge scene labels, or a combination thereof.

Optionally, the method further includes:

for each tourist, collecting video data of the tourist by using a preset camera, and cutting the video data into a plurality of video sections;

for each video segment, extracting user emotional characteristics, and outputting the user emotional characteristics to a pre-trained first neural network model to obtain an emotional index corresponding to the video segment, wherein the emotional characteristics include:

sound, motion, expression or a combination thereof;

when the emotion index is higher than a first preset threshold value, taking the video segment as a target video segment;

and adding the target video segment into the VLOG of the tourist.

Optionally, when the emotional features of the user are extracted, the method further includes:

evaluating the aesthetic feeling index of each video frame of each video segment by utilizing a pre-constructed second neural network;

when the aesthetic feeling index is higher than a second preset threshold value, taking the video frame as a target video frame;

adding the target video frame to the guest's VLOG.

for each video segment, extracting images of tourists contained in the video segment, and evaluating the color index, the stature index and the lap index of the tourists by utilizing a pre-constructed third neural network;

taking an index with a value larger than a third preset threshold value from the color index, the stature index and the lap index as a target index;

when the target index is a color value index, using a video segment of a close shot and/or a video frame of the close shot contained in the video segment;

when the target index is a figure index or a cross-over index, adding a video segment of a distant view and/or a video frame of the distant view included in the video segment into the VLOG of the tourist;

when the target index is a color value index, adding a video segment of a close shot and/or a video frame of the close shot included in the video segment of the close shot into the VLOG of the tourist;

and when the target index is a color value index, a stature index and a cross-matching index, adding a video frame of a close scene in the video segment of the close scene and/or a video frame of a close scene in the video segment of the close scene, and a video frame of a far scene in the video segment of the far scene and/or a video frame of the far scene in the VLOG of the tourist.

Optionally, when the target index does not include the color value index, the stature index and the lap index, extracting a video segment and/or a video frame of the visitor, and adding the extracted content into the VLOG of the visitor.

Optionally, the extracting the video segments and/or the video frames of the guest includes:

evaluating a second aesthetic index of each video frame by utilizing a pre-constructed fourth neural network aiming at each video frame of each video segment; calculating a third aesthetic index of the video segment based on the aesthetic index of the video frame; when the second aesthetic index is larger than a fourth preset threshold value, adding the video frame into the VLOG of the tourist; and when the third aesthetic index is larger than a fifth preset threshold value, adding the video segment into the VLOG of the tourist. Optionally, after extracting the video segments and/or video frames of the guest, before adding the extracted content to the guest's VLOG, the method further comprises:

and performing beautifying processing on the images of the tourists in the extracted content by using a beautifying algorithm.

The invention also provides a scenic spot human-computer interaction device, which comprises:

the system comprises an establishing module, a searching module and a display module, wherein the establishing module is used for establishing a knowledge base aiming at a target scenic spot in advance, and the knowledge base comprises: one or a combination of the playing knowledge, the traffic knowledge, the food culture knowledge and the popular characteristic knowledge in the scenic spot; the acquisition module is used for acquiring a voice ticket-taking instruction sent by the tourist and analyzing the voice ticket-taking instruction of the tourist by using the NLP module, wherein the voice ticket-taking instruction comprises: a ticket fetching instruction; carrying out authentication processing on the tourist and drawing a ticket;

The invention also provides a scenic spot human-computer interaction system, which comprises: the system comprises an input/output controller, an in-place sensor, a human body detection sensor, a help calling button, a ticket gate indicator lamp and a buzzer, wherein the in-place sensor, the human body detection sensor, the help calling button, the ticket gate indicator lamp and the buzzer are connected with the input/output controller;

the intelligent bank card comprises a main controller connected with an input/output controller, a Unionpay card module, a two-dimensional code reader, an identity card reader, a camera, a receipt printer, an entrance ticket printer, a touch display screen, an LED rolling screen and a voice module, wherein the main controller is used for executing the method, and the Unionpay card module, the two-dimensional code reader, the identity card reader, the camera, the receipt printer, the entrance ticket printer, the touch display screen, the LED rolling screen and the voice module are connected with the main controller.

Compared with the prior art, the invention has the following advantages:

according to the embodiment of the invention, through the NLP intelligent answering module, the man-machine interaction of the ticket selling and taking link is realized, the intelligent question answering is realized from voice recognition, the explanation on basic knowledge, basic environment, basic service and the like of scenic spots can be realized, and the participation sense of tourists is enhanced. The ticket selling link has temperature, so that tourists are willing to use the family interaction equipment, the utilization rate of the self-service device is improved, and further the comprehensive benefits of scenic spots are improved.

Drawings

Fig. 1 is a schematic flow chart of a scenic spot human-computer interaction method according to an embodiment of the present invention;

FIG. 2 is a schematic diagram illustrating a method for man-machine interaction in a scenic spot according to an embodiment of the present invention;

fig. 3 is a schematic diagram of an algorithm architecture in a scenic spot human-computer interaction method according to an embodiment of the present invention;

fig. 4 is a schematic diagram of a VLOG generation algorithm in a scenic spot human-computer interaction method according to an embodiment of the present invention;

fig. 5 is a schematic structural diagram of a human-computer interaction system in a scenic spot according to an embodiment of the present invention.

Detailed Description

The following examples are given for the detailed implementation and specific operation of the present invention, but the scope of the present invention is not limited to the following examples.

Example 1

The embodiment of the invention can be deployed on a self-traveling assistant platform.

Fig. 1 is a schematic flow chart of a scenic spot human-computer interaction method according to an embodiment of the present invention; FIG. 2 is a schematic diagram illustrating a method for man-machine interaction in a scenic spot according to an embodiment of the present invention; fig. 3 is a schematic diagram of an algorithm architecture in a scenic spot human-computer interaction method provided by an embodiment of the present invention, as shown in fig. 1-3, the method includes: step S1: pre-establishing a knowledge base aiming at a target scenic spot, wherein the knowledge base comprises: and one or a combination of the playing knowledge, the traffic knowledge, the food culture knowledge and the popular characteristic knowledge in the scenic spot.

Illustratively, the knowledge base constructed for the tourism industry comprises the playing knowledge, the traffic knowledge, the food culture knowledge and the popular characteristic knowledge for scenic spots, and fully covers the information consultation requirements of tourists before, during and after the tourism.

Further, a customized knowledge base can be constructed for different scenes, including:

and (3) conference knowledge: aiming at a knowledge base framework of the exhibition, the common relevant knowledge of the exhibition is included; the scene label is a conference scene label;

research knowledge: a knowledge framework aiming at research scenes comprises research culture and culture content related knowledge; the scene label is a research scene label;

hotel knowledge: the knowledge base framework aiming at hotel and accommodation scenes comprises hotel services, catering services, invoices, check-in and other related knowledge; the scene label is a hotel scene label;

natural encyclopedia knowledge: a knowledge framework for museums and science popularization property scenes, including natural science related common knowledge; the scene label is a natural knowledge scene label;

history culture knowledge: the knowledge frame customized for the historical culture venue comprises local culture and historical humanity related knowledge, and the scene label is a historical culture knowledge scene label.

S2: the method comprises the steps of obtaining a voice ticket-taking instruction sent by a tourist, and analyzing the voice ticket-taking instruction of the tourist by using an NLP (line segment processing) module, wherein the voice ticket-taking instruction comprises the following steps: a ticket fetching instruction; and (4) carrying out authentication processing on the tourists and drawing a ticket. The multifunctional integrated scenic spot marketing device automatically acquires ticket business information of the tourists by means of face recognition, identity card induction or mobile phone numbers of the tourists and the like. The printing module in the multifunctional integrated scenic spot marketing device is started, the ticket of the tourist is made and printed, and then the printing function of the ticket is realized, and the tourist is prompted to carry the ticket to check the ticket. Or, the tourist sends a ticket buying instruction to the multi-function integrated scenic spot marketing device, for example, the device receives the instruction and then starts the NLP module and the self-traveling assistant platform; the natural language is analyzed through the NLP module, and the multifunctional integrated scenic spot marketing device can meet mobile payment modes such as code scanning payment and face brushing payment according to ticket purchasing requirements of tourists; through interacting with the tourists, the printing of paper tickets and the generation of electronic tickets are supported, and the tourists can conveniently select the paper tickets according to own preferences. In addition, voice operation is simpler than clicking operation which requires searching for a specific single page or key.

Specifically, NLP (Natural Language Processing) module research is a comprehensive study, which is far from machine learning algorithms. Text changes are more complex than images or speech, which is a cross domain of computer science, artificial intelligence, and linguistics. The target is to let the computer process or "understand" natural language, and divide task NLP basic technology such as language translation and question answering into three types of lexical analysis, syntactic analysis and semantic analysis. The lexical analysis comprises word segmentation and part-of-speech tagging, wherein the word segmentation method is realized based on a word list matching and statistical model, namely, character strings are scanned word by word, and when a substring of the character strings is found to be the same as words in a word list, the character strings are matched; and modeling the Chinese according to the part of speech and the statistical characteristics of the manual labeling, calculating the probability of the occurrence of various word segmentations through the model, and taking the word segmentation result with the maximum probability as a final result. The purpose of part-of-speech tagging is to assign a specific category to each word in the sentence, i.e. to tag part-of-speech for each word in the word segmentation result. A method based on rules and a method of statistical models, a part-of-speech disambiguation rule is built by a part-of-speech collocation relationship and a context; and calculating the occurrence probability of various parts of speech through the model, and taking the part of speech with the maximum probability as a final result.

Syntactic parsing is the main task of parsing a syntactical parsing, which is to parse an input text sentence (string) to obtain a sentence syntactical structure. On the one hand the nlu (natural language understanding) task's own requirements, and on the other hand may provide support for other nlu tasks. Tasks are classified into 3 types according to different expression forms of syntax structures: dependency syntax analysis dependency syntax parsing, mainly aiming at identifying the interdependence relationship among the vocabularies in the sentence; phrase structure syntax analysis phrase-structure syntax parsing, also called syntactic analysis syntax parsing, the main task is to identify the hierarchical syntax relationship between phrase structures and phrases in sentences; the deep grammar syntactic analysis is mainly used for carrying out deep syntactic and semantic analysis on a sentence by utilizing deep grammars, wherein the deep grammars comprise adjacent lexical trees, functional lexical grammars, combined category grammars and the like.

Semantic analysis refers to meaning included in natural language, and in the field of computer science, semantics can be understood as meaning of a concept represented by a real-world object corresponding to data. Semantic analysis refers to applying various machine learning methods to make a machine learn and understand semantic contents represented by a text. Any understanding of a language can be classified as semantic analysis, involving linguistics, computational linguistics, artificial intelligence, machine learning, and even cognitive language. The ultimate goal of semantic analysis is to understand the true meaning of a sentence expression.

S201 (not shown in the figure): the method comprises the steps of preprocessing an original voice signal input by a tourist and carrying out A/D conversion. Converting the sound continuous waveform into discrete data points by sampling and quantizing with a certain sampling rate and sampling bit number; the MFCC feature extraction is used, and a first-order high-pass filter is used, so that the acoustic model can better utilize a high-frequency formant, and the identification accuracy is improved; endpoint detection and frame windowing, extracting the significant portion of the speech input data. In this MFCC, windowing uses a Hamming window where the edges are smoothly dropped to 0.

S202 (not shown in the figure): and (4) feature processing, namely using Mel cepstrum coefficient feature processing. And extracting MFCC parameters and PLCC parameters, performing maximum likelihood value estimation on a training process of the four parameters of a state transition matrix A, a weight matrix C with mixed Gaussian distribution, a mean vector II and a covariance matrix U, and processing the speech input characteristics of the tourists.

S203 (not shown in the figure): and generating a voice training template, and identifying according to distortion judgment criteria such as Euclidean distance, covariance matrix, Bayesian distance and the like.

S204 (not shown): and performing pattern matching, namely using hidden Markov pattern matching of the HMM to determine an optimal state transition sequence, and calculating the output probability of the observation sequence to the HMM model, or performing judgment according to the speech training template of S103. And judging the recognition result of the voice command of the tourist according to the judgment result.

S205 (not shown): obtaining an identification result, using OpenMP programming to realize multi-core tourist voice parallel operation, using OpenMP to improve two-dimensional discrete fast Fourier transform, using a voice matching result as a multi-operation result, finally programming through an FFT algorithm to output a natural sequence, and outputting a tourist query result.

S3: and acquiring a voice consultation instruction sent by a user, matching the voice consultation instruction based on a pre-established knowledge base, and outputting a matching result.

The tourists can send out consultation instructions to the multi-function integrated scenic spot marketing device according to the demands of the tourists, and the device starts the NLP module to communicate with the tourists after receiving the instructions; the method comprises the steps of analyzing natural languages through an NLP (neural-linear Programming) module, supporting multiple languages such as Mandarin, northeast, Cantonese and English to communicate, matching each scene label according to analyzed semantics to obtain a scene label with the highest matching degree, matching the analyzed semantics with each knowledge in the scene label according to the scene label to obtain a matching result, and displaying the matching result to a user.

For example: the tourists need to consult nearby food, and can say that the device says 'what food is recommended nearby', and the multifunctional integrated scenic spot marketing device lists a food ranking list; when the tourist needs to consult a nearby toilet, the tourist can say that the nearby toilet is located with the device, the device can display a map, show the current position of the tourist, mark the position of the nearby toilet, and meanwhile, the device can lead the tourist to go to the nearby toilet. Furthermore, a user track database, a behavior database, a question and answer database, a basic travel database and the like can be preset in the multifunctional integrated scenic spot marketing device, and the data are input into the tourist question prediction model, the consultation recommendation algorithm model and the guidance recommendation algorithm model, so that the behavior prediction of the tourists is realized, and the experience of the tourists is improved.

According to the embodiment of the invention, through the NLP intelligent answering module, the man-machine interaction of the ticket selling and taking link is realized, the intelligent question answering is realized from voice recognition, the explanation on basic knowledge, basic environment, basic service and the like of scenic spots can be realized, and the participation sense of tourists is enhanced. The temperature is kept in the ticketing link, the utilization rate of the self-service device is improved, and the man-machine interaction in the ticketing link is realized;

moreover, a powerful knowledge base is created, and a closed loop of data acquisition, analysis and feedback is constructed, so that the quality and efficiency of customer service can be continuously and iteratively improved; the data can provide support for management, marketing of the destination platform, and can further be used to enable ecological merchants.

In addition, the cognitive ability is continuously improved by means of a deep learning technology. And providing multiple intelligent text processing capabilities including intelligent word segmentation, entity recognition, keyword extraction, intelligent chatting, text and travel knowledge query and the like. The method can be widely applied to scenes such as tourist answering, function retrieval in the journey, complaint analysis in the journey process and the like, and the text intelligent requirements in the literature and travel industry are met.

Example 2

Fig. 4 is a schematic view of a VLOG generation algorithm in a scenic spot human-computer interaction method provided by an embodiment of the present invention, as shown in fig. 4, based on embodiment 1 of the present invention, embodiment 2 adds the following steps on the basis of embodiment 1:

s4 (not shown): and aiming at each tourist, the video data of the tourist is collected by utilizing a preset camera, and the video data is cut into a plurality of video sections.

The cameras are configured according to needs to carry out point location deployment; the intelligent edge all-in-one machine is locally deployed, the number of the intelligent edge all-in-one machines is configured according to the number of the cameras, the intelligent edge all-in-one machines are generally deployed in a central machine room, are in data transmission with a camera local area network, and are in butt joint with a VLOG operation cloud platform through a public network fixed IP. The VLOG operation platform is deployed in the Alice cloud, a scenic spot administrator can log in and manage through a web terminal, and a consumer can log in and select to manufacture and obtain the VLOG through a small program and an APP. The system automatically cuts the guest's video into video segments, typically 3-5 seconds in length.

S5 (not shown): for each video segment, extracting user emotional characteristics, and outputting the user emotional characteristics to a pre-trained first neural network model to obtain an emotional index corresponding to the video segment, wherein the emotional characteristics include: sound, motion, expression or a combination thereof. The user expression may include: fear, surprise, feeling, embarrassment, shyness, happiness, etc.

It can be understood that the training method of the first neural network model includes: manually identifying volume data, language data and the like in sound generated by sample tourists in the scenic spot in the playing process to obtain an emotion index, and then taking the emotion index as a tag of the sound data to obtain a sound sample; manually identifying according to action amplitude, action acceleration and the like in the actions of the sample tourists, identifying the exciting degree of the sample tourists to further obtain an emotion index, and taking the emotion index as a label of the action data of the sample tourists to further obtain an action sample of the sample tourists; similarly, quantification is carried out according to the degree of the expression, such as the level of surprise of the sample tourist and the level of embarrassment of the sample tourist, so as to obtain a corresponding emotion index and further obtain an expression sample; and training the first neural network model by using the sound sample, the action sample and the expression sample to further obtain a trained model.

It can be understood that when the game of the tourist is finished and the exclusive tourist plays the photos and videos, the privacy and the uniqueness of the photos need to be ensured.

S6 (not shown): and when the emotion index is higher than a first preset threshold value, taking the video segment as a target video segment.

It is understood that the deeper the emotional depth, the greater the value corresponding to the emotional index. For example, a larger voice indicates a higher emotion index, and a faster pace indicates a higher emotion index. The more embarrassing the expression of the tourist, the higher the emotion index is.

Therefore, in the embodiment of the invention, the video segment with the emotion index ranking of the top 10% can be selected as the target video segment; and taking a boundary of the top 10% of the ranking as a first preset threshold.

S7 (not shown): and adding the target video segment into the VLOG of the tourist.

Furthermore, when the photos of the tourists are synthesized into the exclusive and private customized VLOG, the images, the color distribution of the videos, the texture characteristics and the shapes of main objects can be identified by using an image identification algorithm, spots are identified by using a Gaussian Laplacian operator detection method (LOG), local characteristics of corner points are identified by using a Harris algorithm and a FAST algorithm, and different characteristics of the images are classified according to different identification algorithms. The guest may select a favorite category or categories based on the classification results. Then, the existing VLOG generation algorithm is used to perform overlay synthesis and arrangement processing on the contents selected by the guest, so as to obtain an arrangement picture, and the arrangement picture processing is called as VLOG.

After the visitor went into the garden, VLOG terminal equipment can snatch the wonderful of visitor in the scenic spot in the twinkling of an eye to upload data information to back-end system platform, when the visitor needed the exclusive video of VLOG oneself, can consult multi-functional integrated scenic spot marketing device. For example, "help me generate my exclusive micro video", the device can find the picture or video information of the corresponding guest by face recognition of the guest, automatically clip and synthesize the picture or video information, and automatically generate the exclusive micro video of the corresponding guest through the VLOG platform.

By means of the VLOG technology, wonderful images and videos of tourists in scenic spots are captured, permanent memory of the tourists in the scenic spots is synthesized, the drips played by the tourists in the scenic spots are recorded, the memory feeling of the tourists to the scenic spots is enhanced, the scenic spots can be circularly played by utilizing the synthesized videos to conduct scenic spot propaganda, and a scenic spot propaganda content library is formed.

The ticket selling mode with single function of the traditional self-service ticket selling and taking machine is enriched, the pressure of queuing for buying tickets in an artificial ticket selling window is relieved, and the working efficiency of selling and taking tickets is improved; the function of interactive communication with the tourists is added, more ways for the tourists to consult problems are provided, the tourists can speak the needs of the tourists, and the operation is more convenient and fast; the wonderful moment of the tourists in the scenic spot is recorded, and a special micro video is generated, so that the satisfaction degree of the tourists is improved; through the circulation of the scenic spot micro-video, on the other hand, propaganda ways of the scenic spot are increased, and the brand awareness of the scenic spot is improved.

Example 3

The embodiment 3 of the invention is added with the following steps on the basis of the embodiment 2:

evaluating the aesthetic feeling index of each video frame of each video segment by utilizing a pre-constructed second neural network; when the aesthetic feeling index is higher than a second preset threshold value, taking the video frame as a target video frame; adding the target video frame to the guest's VLOG.

A large number of video segments or video frames are stored in the system, and the aesthetic feeling of the video segments or the video frames is subjective manually from the aspects of composition, light and shadow cooperation, coordination between people and nature and the like; and (3) calculating an average value by adopting multiple scoring times by multiple people, and taking the average value as the aesthetic indexes of the video frames or video segments, wherein the value range of each index is 1-10, and the aesthetic indexes are no exception. And then training a second neural network by using the scored video frames and video segments, and then performing aesthetic evaluation on the video frames or the video segments of the tourists by using the trained second neural network to obtain a corresponding aesthetic index.

By applying the embodiment of the invention, pictures with high aesthetic feeling can be screened out and added into the VLOG, and because the pictures are shot by the camera, on one hand, the pictures can give surprise to tourists in unconscious times, and on the other hand, the invention uses the second neural network to identify the aesthetic feeling pictures, so that the pictures are more professional, the effect of the obtained pictures is better, and the user experience is further improved.

Or, when the emotional features of the user are extracted, the method further includes:

taking an index with a value larger than a third preset threshold value from the color index, the stature index and the lap index as a target index; when the target index is a color value index, using a video segment of a close shot and/or a video frame of the close shot contained in the video segment; when the target index is a figure index or a cross-over index, adding a video segment of a distant view and/or a video frame of the distant view included in the video segment into the VLOG of the tourist; when the target index is a color value index, adding a video segment of a close shot and/or a video frame of the close shot included in the video segment of the close shot into the VLOG of the tourist; and when the target index is a color value index, a stature index and a cross-matching index, adding a video frame of a close scene in the video segment of the close scene and/or a video frame of a close scene in the video segment of the close scene, and a video frame of a far scene in the video segment of the far scene and/or a video frame of the far scene in the VLOG of the tourist.

For example, the system makes intelligent recommendations based on the guest's color value, stature, fit, make-up.

If the visitor has a high color value and a good stature and wins the lottery, the VLOG is generated by fusing a far scene and a near scene and a video and a photo.

The VLOG is generated in an intelligent mode by mainly selecting distant view or image conversion processing due to high color value and common stature of tourists. The tourist has common color value and good stature. The lens special effect and the beauty special effect are started, and the visitor color value is improved comprehensively. Intelligent video compositing uses mainly perspective to generate VLOG.

And when the visitor image is not good in the color value, the figure and the putting on and putting off, performing the beautifying processing on the visitor image in the extracted content by using the existing beautifying algorithm.

Or when the visitor is not good in color value, stature and putting on, the standard is properly reduced, such as:

regarding each video frame of each video segment, taking the video frame with the aesthetic feeling index ranking 10 percent as the video frame of the tourist;

and taking the video segment with the aesthetic index ranking 10 percent of the video segments as the video segment for the tourist.

Therefore, by applying the embodiment of the invention, the condition that the video frames or video segments of the tourists do not meet the common aesthetic standard can be avoided, a certain number of video frames and video segments can still be screened out, and the experience of the tourists is improved.

Example 4

Based on any one of embodiments 1 to 3, embodiment 4 of the present invention provides a scenic spot human-computer interaction device, including:

Example 5

Fig. 5 is a schematic structural diagram of a human-computer interaction system in a scenic spot according to an embodiment of the present invention, and as shown in fig. 5, based on any one of embodiments 1 to 4, embodiment 5 of the present invention provides a human-computer interaction system in a scenic spot, where the system includes: the system comprises an input/output controller, an in-place sensor, a human body detection sensor, a help calling button, a ticket gate indicator lamp and a buzzer, wherein the in-place sensor, the human body detection sensor, the help calling button, the ticket gate indicator lamp and the buzzer are connected with the input/output controller;

a main controller connected with the input and output controller, the main controller is used for executing the method as claimed in any one of claims 1 to 8, and a Unionpay card module, a two-dimensional code reader, an identity card reader, a camera, a receipt printer, an entrance ticket printer, a touch display screen, an LED scroll screen and a voice module which are connected with the main controller. The device adopts an I/O controller designed by an embedded industrial control computer and a microprocessor, has definite division of labor, strong function, high reliability and convenient function expansion; the damp-proof, mould-proof and salt fog-proof design of the electronic circuit board ensures the long-term stable operation of the I/O controller of the self-service machine; strict grounding measures and lightning protection designs are adopted, and the safety is good; standard components and a universal interface are adopted, so that upgrading and maintenance are convenient;

TABLE 1 multifunctional integrated scenic spot marketing device VLOG module function list

TABLE 2 parameters of multifunctional integrated scenic spot marketing device

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims

1. A scenic spot human-computer interaction method is characterized by comprising the following steps:

2. The human-computer interaction method for scenic spots as claimed in claim 1, wherein the knowledge base of the target scenic spot further comprises a customized knowledge base, and the establishment process of the customized knowledge base comprises:

3. A scenic spot human-computer interaction method as claimed in claim 1, further comprising:

for each video segment, extracting user emotional characteristics, and outputting the user emotional characteristics to a pre-trained first neural network model to obtain an emotional index corresponding to the video segment, wherein the emotional characteristics include: sound, motion, expression or a combination thereof;

and adding the target video segment into the VLOG of the tourist.

4. A scenic spot human-computer interaction method according to claim 3, wherein when the emotional features of the user are extracted, the method further comprises:

adding the target video frame to the guest's VLOG.

5. The human-computer interaction method for scenic spots as claimed in claim 1, wherein when the emotional features of the user are extracted, the method further comprises:

6. The human-computer interaction method for scenic spots of claim 5, wherein when none of the target indexes comprises a color value index, a stature index and a cross-matching index, video segments and/or video frames of the tourists are extracted, and the extracted contents are added to VLOGs of the tourists.

7. A scenic spot human-computer interaction method according to claim 6, wherein the extracting of video segments and/or video frames of tourists comprises:

8. A scenic spot human-computer interaction method as claimed in claim 6, wherein after extracting a guest's video segment and/or video frame, before adding the extracted content to the guest's VLOG, the method further comprises:

9. A scenic spot human-computer interaction device, the device comprising:

the system comprises an establishing module, a searching module and a display module, wherein the establishing module is used for establishing a knowledge base aiming at a target scenic spot in advance, and the knowledge base comprises: one or a combination of the playing knowledge, the traffic knowledge, the food culture knowledge and the popular characteristic knowledge in the scenic spot;

the acquisition module is used for acquiring a voice ticket-taking instruction sent by the tourist and analyzing the voice ticket-taking instruction of the tourist by using the NLP module, wherein the voice ticket-taking instruction comprises: a ticket fetching instruction; carrying out authentication processing on the tourist and drawing a ticket;

10. A scenic spot human-computer interaction system, the system comprising: the system comprises an input/output controller, an in-place sensor, a human body detection sensor, a help calling button, a ticket gate indicator lamp and a buzzer, wherein the in-place sensor, the human body detection sensor, the help calling button, the ticket gate indicator lamp and the buzzer are connected with the input/output controller;

a main controller connected with the input and output controller, the main controller is used for executing the method as claimed in any one of claims 1 to 8, and a Unionpay card module, a two-dimensional code reader, an identity card reader, a camera, a receipt printer, an entrance ticket printer, a touch display screen, an LED scroll screen and a voice module which are connected with the main controller.