WO2021227805A1 - 地下空间品质评价及其可视化呈现方法及系统 - Google Patents

地下空间品质评价及其可视化呈现方法及系统 Download PDF

Info

Publication number
WO2021227805A1
WO2021227805A1 PCT/CN2021/088812 CN2021088812W WO2021227805A1 WO 2021227805 A1 WO2021227805 A1 WO 2021227805A1 CN 2021088812 W CN2021088812 W CN 2021088812W WO 2021227805 A1 WO2021227805 A1 WO 2021227805A1
Authority
WO
WIPO (PCT)
Prior art keywords
underground space
picture
pictures
score
location
Prior art date
Application number
PCT/CN2021/088812
Other languages
English (en)
French (fr)
Inventor
雷升详
李文胜
李庆
周彪
桂颖彬
谢雄耀
梁田
王华兵
贠毓
许洋
尹巧
Original Assignee
中铁第四勘察设计院集团有限公司
同济大学
中国铁建股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中铁第四勘察设计院集团有限公司, 同济大学, 中国铁建股份有限公司 filed Critical 中铁第四勘察设计院集团有限公司
Publication of WO2021227805A1 publication Critical patent/WO2021227805A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B25HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
    • B25JMANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
    • B25J19/00Accessories fitted to manipulators, e.g. for monitoring, for viewing; Safety devices combined with or specially adapted for use in connection with manipulators
    • B25J19/02Sensing devices
    • B25J19/021Optical sensing devices
    • B25J19/023Optical sensing devices including video camera means
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B25HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
    • B25JMANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
    • B25J5/00Manipulators mounted on wheels or on carriages
    • B25J5/005Manipulators mounted on wheels or on carriages mounted on endless tracks or belts
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B25HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
    • B25JMANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
    • B25J9/00Programme-controlled manipulators
    • B25J9/0009Constructional details, e.g. manipulator supports, bases
    • B25J9/0027Means for extending the operation range
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/045Combinations of networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W4/00Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02Services making use of location information
    • H04W4/025Services making use of location information using location based information parameters
    • H04W4/026Services making use of location information using location based information parameters using orientation information, e.g. compass
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20081Training; Learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20084Artificial neural networks [ANN]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30168Image quality inspection
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30204Marker

Definitions

  • This application relates to the technical field of space quality evaluation, and in particular to a method and system for underground space quality evaluation and its visual presentation.
  • the platform related to spatial perception is mainly MIT's Place Pulse, which intercepts Google Street View pictures and the corresponding coordinates of the pictures through the crawler algorithm, places the pictures on the web platform, and collects the scoring data of the online users on the pictures.
  • Place Pulse has major limitations. First of all, because its image acquisition comes from Google Street View images, this also means that it cannot reach the underground space, and no one currently provides the corresponding space images in the underground space.
  • Place Pulse’s scoring data collection method is carried out through a public web platform, which does not have any reward mechanism. There is not enough data that can be involved, resulting in insufficient scoring accuracy.
  • the embodiments of the present application propose a method and system for evaluating underground space quality and its visual presentation.
  • the embodiments of the present application provide an underground space quality evaluation and a visual presentation method thereof, including the following steps:
  • the embodiments of the present application also provide an underground space quality evaluation and a visual presentation system for implementing the above method, including a picture collection robot and an underground space intelligent perception platform;
  • the picture collection robot is used to collect pictures in multiple directions of various locations in the existing underground space;
  • the underground space intelligent perception platform includes:
  • the data collection module is configured to put the collected pictures into the data collection platform, and collect the results of the evaluation of the four indicators of the space brightness, space comfort, space richness, and environmental artistry of each picture by network users through the crowdsourcing mechanism. And analyze the score data of each picture;
  • the quality evaluation module is configured to train through the deep learning algorithm combined with the score data and feature information of each picture to obtain a model for inferring the picture score data based on the picture characteristics; use the trained model to evaluate the number of locations in the underground space that need to be evaluated. Score the pictures in each location, and use the score data of the pictures in multiple locations in each location to obtain the score value of each location;
  • the visual presentation module is configured to mark each location on the underground space plan with different colors according to the score of each location to form a planar distribution map of the score result and display it.
  • the underground space quality evaluation and its visual presentation method and system provided by the embodiments of the present application use a picture collection robot to collect pictures of multiple orientations of various locations in the existing underground space, so as to solve the current problem of lack of pictures in underground space through crowdsourcing.
  • the mechanism collects the scoring data of the image and trains it with a deep learning algorithm to obtain a model that infers the image scoring data based on the characteristics of the image.
  • Using the model to score the image can improve the accuracy of the scoring and facilitate the evaluation.
  • By using different colors according to the level of the score of each location Mark the locations on the underground space plan to form a plan distribution map of the score results and display them, so as to realize the visual presentation of the evaluation results and facilitate users to obtain the information of the evaluation results.
  • the embodiment of the application provides corresponding image data of underground space, which can provide image data for corresponding underground space navigation in the future, and the collected scoring data can provide a data basis for related work of quantitative evaluation of underground space quality.
  • Fig. 1 is a flowchart of an underground space quality evaluation and its visual presentation method provided by an embodiment of the application;
  • FIG. 2 is a block diagram of an underground space quality evaluation and its visual presentation system provided by an embodiment of the application
  • FIG. 3 is a schematic diagram of the structure of an image acquisition robot provided by an embodiment of the application.
  • Figure 4 is a schematic diagram of a lifting mechanism provided by an embodiment of the application.
  • Fig. 5 is a control principle diagram of an image acquisition robot provided by an embodiment of the application.
  • an embodiment of the present application provides an underground space quality evaluation and a visual presentation method thereof, which can be applied to an underground space intelligent perception platform.
  • the method includes the following steps:
  • the picture collection robot travels at a predetermined speed and stops moving when it reaches a certain distance.
  • the picture is taken, and the picture collection robot continues to move after the picture is taken until the picture is taken in each location.
  • the collected picture groups corresponding to each location are numbered, and two photos of the same orientation at different locations or two photos of the same location with different orientations are put into the data collection platform for display each time, and the reward mechanism is used.
  • the data collection method collects the results of the comparison and selection of the two pictures given by the network users under the preset questions about the brightness of the space, the comfort of the space, the richness of the space, and the artisticness of the environment.
  • the comparison of the two pictures can be used in the form of (Xi, Xj, Y).
  • Xi and Xj represent two pictures respectively, and Y is 0 or 1.
  • the TrueSkill algorithm is used to sort the pictures to obtain the score data of each picture.
  • the TrueSkill algorithm is a combination of the Elo ranking method and the Bayesian rule, specifically: assign a normal distribution to each object, and the average value represents the true Ability, variance represents the uncertainty of the system's guess of the user's true ability. At the beginning, it is assumed that the mean and variance of the distribution of each object are the same, and then the data is used to continuously update the distribution of each object.
  • the data collection method with reward mechanism is for example: (1) A tree planting game that can be combined with a small program. The more questions answered, the more nutrients the tree can obtain. (2) Use questions as a means of resurrection for a certain game, by answering questions and then resurrecting. (3) Regard the question as a verification question for a certain web interface login. (4) Make the question into a psychological test.
  • the semantic features and color features of each picture are extracted, and the full convolutional neural network (FCN) is used to train the score data, semantic features and color features of each picture obtained, and the results can be inferred based on the semantic features and color features of the pictures.
  • FCN full convolutional neural network
  • the picture collection robot will collect pictures and coordinates of various locations of the underground space site. If the underground space site that needs to be evaluated has not been built, it will be The BIM (Building Information Modeling) model of the space takes screenshots of various locations in multiple directions and records the coordinates of the screenshot locations.
  • BIM Building Information Modeling
  • the method of collecting pictures of multiple orientations of various locations in the underground space site by a picture collection robot is the same as the above-mentioned method of collecting pictures of multiple orientations of various locations in the existing underground space by a picture collection robot, and will not be repeated here.
  • the method for collecting pictures and coordinates of multiple orientations of various locations in the underground space site through a picture collection robot includes:
  • the positioning base station adopts ceiling-mounted or flat-panel positioning base stations, which can be installed on the top of the underground space or on the top of the column.
  • the image collection robot passes Obtain the signals of at least two base stations distributed in the underground space, determine the signal angle of each base station, determine the current coordinates of the robot relative to the base station through the AOA (angle of arrival) algorithm, and combine the absolute coordinates of the base station map and the picture to collect the robot Calculate relative to the coordinates of the base station to obtain the absolute coordinates of the current location of the robot, that is, the coordinates of the corresponding location of the underground space site.
  • each location on the underground space plan map is marked with different colors according to the score of each location to form a flat distribution map of the score results.
  • Each question corresponds to an indicator of spatial brightness, spatial comfort, spatial richness, and environmental artistry, and corresponds to a flat distribution map of the scoring results.
  • the flat distribution map of the scoring results is displayed, and the user can click the flat distribution map of the scoring results Get the score value of each location on each location on the.
  • the above-mentioned related pictures and scoring data can also be opened to users, and different researchers can download them according to their own needs.
  • This kind of underground space quality evaluation and its visual presentation method uses a picture collection robot to collect pictures of various locations in the existing underground space to solve the current problem of lack of pictures in underground space and collect through crowdsourcing mechanism.
  • the scoring data of the picture is trained by deep learning algorithms to obtain a model that infers the scoring data of the picture based on the characteristics of the picture.
  • the model is used to score the picture, which can improve the accuracy of the scoring and facilitate the evaluation.
  • Each location on the spatial plan is marked to form a plan distribution map of the scoring results and displayed, so as to realize the visual presentation of the evaluation results and facilitate users to obtain the information of the evaluation results.
  • the embodiment of the application provides corresponding image data of underground space, which can provide image data for corresponding underground space navigation in the future, and the collected scoring data can provide a data basis for related work of quantitative evaluation of underground space quality.
  • the underground space quality evaluation and its visual presentation method include the following steps in sequence; (1), the robot collects space pictures; (2), puts the pictures on the data collection platform; (3) ), crowdsourcing mechanism to compare pictures; (4), using a ranking algorithm to obtain scoring data according to the sorting of pictures; among them, the ranking algorithm can be TrueSkill algorithm; (5), using deep learning to combine scoring data and semantic features, color features for training; (6) Determine whether the model meets the requirements; if yes, proceed to step (7); if not, repeat steps (1) to (6) in sequence until the model meets the requirements; (7), determine whether the site is built; if yes , Proceed to step (8); if not, collect site pictures and coordinates through the robot; (8), program screenshots or manual screenshots according to the designed BIM model, and record the coordinates; (9), perform site pictures that need to be evaluated Evaluation; (10), to form a flat distribution map of the scoring results.
  • an embodiment of the present application also provides an underground space quality evaluation and its visual presentation system, which is used to implement the above method, and the system includes a picture collection robot and an underground space intelligent perception platform 2;
  • the picture collection robot is used to collect pictures in multiple directions of various locations in the existing underground space;
  • the underground space intelligent perception platform 2 includes:
  • the data collection module 21 is configured to put the collected pictures into the data collection platform, and collect the results of the evaluation of the four indicators of space brightness, space comfort, space richness, and environmental artistry of each picture by network users through a crowdsourcing mechanism , And analyze the score data of each picture;
  • the quality evaluation module 22 is configured to train by combining the score data and feature information of each picture through a deep learning algorithm to obtain a model for inferring the picture score data based on the picture features; Score the pictures in multiple directions, and use the score data of the pictures in multiple directions in each location to obtain the score value of each location;
  • the visual presentation module 23 is configured to mark each location on the underground space plan map with different colors according to the score level of each location to form a planar distribution map of the score result and display it.
  • the picture collection robot 1 includes a housing, a central processing unit 11, a rotating lifting device, a photographing device 13, and a moving device 16.
  • the central processing unit 11 is located in the housing, and the moving The device 16 is fixed on both ends of the housing, the rotary lifting device is fixed above the housing, the photographing device 13 is fixed on the rotary lifting device, the moving device 16, the rotary lifting device, the photographing device
  • the devices 13 are electrically connected to the central processing unit 11.
  • the central processing unit 11 controls the mobile device 16 to stop moving, and at the same time starts the rotary lifting device and the photographing device 13, each time the rotary lifting device rises to the distance of 1.70m from the ground to the photographing device 13 to simulate a human the height of.
  • the photographing device 13 includes dual cameras, and the distance between the two cameras is the distance between human eyes, which simulates the space seen by the human eyes.
  • the moving device 16 adopts crawler belt transmission, and moves smoothly.
  • the rotary lifting device includes a lifting mechanism 12 and a rotating mechanism fixed on the lifting mechanism 12, and the rotating mechanism includes a horizontal rotating block 14 that rotates horizontally relative to the lifting mechanism 12 and
  • the rotating block 14 is a vertical rotating block 15 that rotates vertically.
  • the photographing device 13 is fixed to the vertical rotating block 15, and the photographing device 13 is raised and lowered by the lifting mechanism 12, and the horizontal direction of the photographing device 13 is realized by the horizontal rotating block 14
  • the vertical rotation of the shooting device 13 is realized by the vertical rotation block 15, so that pictures of various locations in the underground space can be taken.

Abstract

一种地下空间品质评价及其可视化呈现方法及系统,该方法包括以下步骤:通过图收集机器人收集现有地下空间的图片;将收集的图片投入到数据收集平台中,通过众包机制收集各图片的空间明暗度、空间舒适性、空间丰富度、环境艺术性四个指标的评价结果;通过深度学习算法得到用于根据图片特征推断图片评分数据的模型;重复进行上述步骤,直至模型的准确度达到要求;获取需要评价的地下空间的图片以及坐标,利用训练好的模型对各图片进行评分;按评分高低用不同颜色对地下空间平面图上的各个地点进行标注形成评分结果平面分布图并进行显示。

Description

地下空间品质评价及其可视化呈现方法及系统
相关申请的交叉引用
本申请基于申请号为202010386869.X、申请日为2020年05月09日的中国专利申请提出,并要求该中国专利申请的优先权,该中国专利申请的全部内容在此以引入方式并入本申请。
技术领域
本申请涉及空间品质评价技术领域,尤其涉及一种地下空间品质评价及其可视化呈现方法及系统。
背景技术
目前与空间感知相关的平台主要以MIT的Place Pulse为主,通过爬虫算法截取到谷歌街景图片及图片相应的坐标,将图片放在网页平台上,收集网络用户对图片的评分数据。但是Place Pulse具有较大的局限性,首先,由于它的图片获取来自于谷歌街景图片,这也意味着它无法触及到地下空间,目前还没人提供地下空间内的相应的空间图片。其次,Place Pulse的评分数据收集方式是通过公开网页平台进行的,是不具有任何的奖励性机制的收集方式,所能涉及到的数据不够多,导致评分不够准确。
鉴于此,有必要提出一种地下空间品质评价及其可视化呈现方法及系统。
发明内容
为解决现有技术存在的上述技术问题,本申请实施例提出一种地下空间品质评价及其可视化呈现方法及系统。
本申请实施例是这样实现的:
一方面,本申请实施例提供一种地下空间品质评价及其可视化呈现方法,包括以下步骤:
S1、通过图收集机器人收集现有地下空间各个地点的多个方位的图片;
S2、将收集的图片投入到数据收集平台中,通过众包机制收集网络用户对各图片的空间明暗度、空间舒适性、空间丰富度、环境艺术性四个指标进行评价的结果,并分析得到各图片的评分数据;
S3、通过深度学习算法结合各图片的评分数据以及特征信息进行训练,得到用于根据图片特征推断图片评分数据的模型;
S4、重复进行步骤S1~S3,直至模型的准确度达到要求;
S5、获取需要评价的地下空间各个地点的多个方位的图片以及坐标,利用训练好的模型对各图片进行评分,并利用各个地点的多个方位的图片的评分数据得到各个地点的评分值;
S6、按各个地点的评分高低用不同颜色对地下空间平面图上的各个地点进行标注形成评分结果平面分布图并进行显示。
另一方面,本申请实施例还提供一种地下空间品质评价及其可视化呈现系统,用于实现上述的方法,包括图片收集机器人以及地下空间智能感知平台;
所述图片收集机器人用于收集现有地下空间各个地点的多个方位的图片;
所述地下空间智能感知平台包括:
数据收集模块,配置为将收集的图片投入到数据收集平台中,通过众包机制收集网络用户对各图片的空间明暗度、空间舒适性、空间丰富度、环境艺术性四个指标进行评价的结果,并分析得到各图片的评分数据;
品质评价模块,配置为通过深度学习算法结合各图片的评分数据以及 特征信息进行训练,得到用于根据图片特征推断图片评分数据的模型;利用训练好的模型对需要评价的地下空间各个地点的多个方位的图片进行评分,并利用各个地点的多个方位的图片的评分数据得到各个地点的评分值;
可视化呈现模块,配置为按各个地点的评分高低用不同颜色对地下空间平面图上的各个地点进行标注形成评分结果平面分布图并进行显示。
本申请实施例提供的这种地下空间品质评价及其可视化呈现方法及系统,通过图片收集机器人收集现有地下空间各个地点的多个方位的图片,解决目前地下空间图片缺失的问题,通过众包机制收集图片的评分数据并通过深度学习算法训练得到根据图片特征推断图片评分数据的模型,利用模型对图片进行评分,可以提高评分的准确度且评价方便,通过按各个地点的评分高低用不同颜色对地下空间平面图上的各个地点进行标注形成评分结果平面分布图并进行显示,实现评价结果的可视化呈现,方便用户获取评价结果信息。本申请实施例提供了相应的地下空间的图片数据,可以为以后相应的地下空间导航提供图片数据,所收集到的评分数据可以为地下空间品质量化评价的相关工作提供数据基础。
附图说明
图1为本申请实施例提供的一种地下空间品质评价及其可视化呈现方法的流程图;
图2为本申请实施例提供的一种地下空间品质评价及其可视化呈现系统的方框图;
图3为本申请实施例提供的图像采集机器人的结构示意图;
图4为本申请实施例提供的升降机构的示意图;
图5为本申请实施例提供的图像采集机器人的控制原理图。
附图标记说明:1-图像采集机器人、11-中央处理单元、12-升降机构、13-拍摄装置、14-水平旋转块、15-竖向旋转块、16-移动装置、2-地下空间 智能感知平台、21-数据收集模块、22-品质评价模块、23-可视化呈现模块。
具体实施方式
下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例仅仅是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其它实施例,都属于本申请保护的范围。
如图1所示,本申请实施例提供一种地下空间品质评价及其可视化呈现方法,可应用在地下空间智能感知平台中,该方法包括以下步骤:
S1、通过图收集机器人收集现有地下空间各个地点的多个方位的图片;
具体地,首先为图片收集机器人设定路径,并设定拍摄间距,图片收集机器人以预定速度行驶,达到一定距离时停止移动,对该地点的前、后、左、右、上五个方位的照片进行拍摄,拍摄完成后图片收集机器人继续移动,直至完成各个地点的图片拍摄。
S2、将收集的图片投入到数据收集平台中,通过众包机制收集网络用户对各图片的空间明暗度、空间舒适性、空间丰富度、环境艺术性四个指标进行评价的结果,并分析得到各图片的评分数据。
具体地,将收集的与各个地点对应的图片组进行编号,每次将两张不同地点的相同方位照片或者两张相同地点不同方位的照片投入到数据收集平台中进行显示,利用具有奖励机制的数据收集方式,收集网络用户在预设的关于空间明暗度、空间舒适性、空间丰富度、环境艺术性的问题下,对所给出的两张图片进行对比选择的结果。可将两两图片的对比用(Xi,Xj,Y)的形式表,Xi和Xj分别表示两张图片,Y为0或1;如果Y为0,则表示Xi在该指标要差于Xj,如果Y为1则表示Xi在该指标要优于Xj。然后采用TrueSkill algorithm根据图片排序得到各图片的评分数据, 其中,TrueSkill algorithm(TrueSkill算法)是Elo排名方法与贝叶斯规则的结合,具体为:给每一个对象分配一个正态分布,均值代表真实能力,方差代表系统对该用户真实能力猜测的不确定程度。一开始假设每个对象分布的均值和方差一致,此后利用数据不断更新每个对象的分布。
所述具有奖励机制的数据收集方式例如:(1)一款可以与小程序结合的种树小游戏,所回答的问题越多,小树获得的养分越多。(2)将问题作为某款游戏的复活手段,通过回答问题然后复活。(3)将问题作为某网页界面登录的验证问题。(4)将问题做成心理测试题。通过具有奖励机制的数据收集方式,可以收集大量的评价数据,减少数据收集时间,并提高评价的准确性。
S3、通过深度学习算法结合各图片的评分数据以及特征信息进行训练,得到用于根据图片特征推断图片评分数据的模型。
具体地,提取各图片的语义特征和颜色特征,采用全卷积神经网络(FCN)对所得到的各图片的评分数据以及语义特征和颜色特征进行训练,得到可以根据图片语义特征和颜色特征推断图片评分数据的模型。
S4、重复进行步骤S1~S3,直至模型的准确度达到要求;这样通过模型可以获取准确度较高的图片评价结果。
S5、获取需要评价的地下空间各个地点的多个方位的图片以及坐标,利用训练好的模型对各图片进行评分,并利用各个地点的多个方位的图片的评分数据得到各个地点的评分值。
具体地,如果需要评价的地下空间场地已经建好,则通过图片收集机器人收集该地下空间场地各个地点的多个方位的图片及坐标,如果需要评价的地下空间场地未建好,则根据该地下空间的BIM(Building Information Modeling,建筑信息模型)模型进行各个地点的多个方位的截图并记录截图地点的坐标。
所述通过图片收集机器人收集该地下空间场地各个地点的多个方位的图片的方法与上述通过图收集机器人收集现有地下空间各个地点的多个方位的图片的方法一致,在此不再赘述。
可选地,所述通过图片收集机器人收集该地下空间场地各个地点多个方位的图片及的坐标的方法包括:
提前在地下空间场地布置至少两个UWB(ultra wide band;超宽带)定位基站,定位基站采用吸顶式或者平板式的定位基站,可安装在地下空间场地顶部或者柱顶部区域,图片收集机器人通过获取地下空间中分布的至少两个基站的信号,判断各个基站的信号角度,通过AOA(angle of arrival,到达角度)算法确定机器人目前相对于基站的坐标,结合基站的地图绝对坐标和图片收集机器人相对于基站的坐标进行计算,获得机器人当前所在位置的绝对坐标,即为该地下空间场地对应地点的坐标。
S6、按各个地点的评分高低用不同颜色对地下空间平面图上的各个地点进行标注形成评分结果平面分布图并进行显示。
具体地,将各个地点的评分值结合之前收集的各个地点的坐标,针对不同的问题,按各个地点的评分高低用不同颜色对地下空间平面图上的各个地点进行标注,形成评分结果平面分布图,每个问题对应空间明暗度、空间舒适性、空间丰富度、环境艺术性中的一个指标,并对应一个评分结果平面分布图,将评分结果平面分布图进行显示,用户可通过点击评分结果平面分布图上的各个地点获取各个地点的评分值。
根据实际需要还可以向用户开放上述相关图片及评分数据,不同研究人员可以根据自身需求进行下载。
本申请实施例提供的这种地下空间品质评价及其可视化呈现方法,通过图片收集机器人收集现有地下空间各个地点的多个方位的图片,解决目前地下空间图片缺失的问题,通过众包机制收集图片的评分数据并通过深 度学习算法训练得到根据图片特征推断图片评分数据的模型,利用模型对图片进行评分,可以提高评分的准确度且评价方便,通过按各个地点的评分高低用不同颜色对地下空间平面图上的各个地点进行标注形成评分结果平面分布图并进行显示,实现评价结果的可视化呈现,方便用户获取评价结果信息。本申请实施例提供了相应的地下空间的图片数据,可以为以后相应的地下空间导航提供图片数据,所收集到的评分数据可以为地下空间品质量化评价的相关工作提供数据基础。
在一个可选实施例中,如图1所示,地下空间品质评价及其可视化呈现方法依次包括以下步骤;(1)、机器人收集空间图片;(2)、将图片投入数据收集平台;(3)、众包机制对比图片;(4)、采用排名算法根据图片排序得到评分数据;其中,排名算法可以为TrueSkill算法;(5)、采用深度学习结合评分数据及语义特征、颜色特征进行训练;(6)、判断模型是否达到要求;若是,则进行步骤(7);若否,则依次重复步骤(1)至(6),直至模型达到要求;(7)、判断场地是否建好;若是,则进行步骤(8);若否,则通过机器人收集场地图片及坐标;(8)、根据设计的BIM模型进行编程截图或者人工截图,记录坐标;(9)、对需要评价的场所图片进行评价;(10)、形成评分结果平面分布图。
如图2所示,本申请实施例还提供一种地下空间品质评价及其可视化呈现系统,用于实现上述的方法,该系统包括图片收集机器人以及地下空间智能感知平台2;
所述图片收集机器人用于收集现有地下空间各个地点的多个方位的图片;
所述地下空间智能感知平台2包括:
数据收集模块21,配置为将收集的图片投入到数据收集平台中,通过众包机制收集网络用户对各图片的空间明暗度、空间舒适性、空间丰富度、 环境艺术性四个指标进行评价的结果,并分析得到各图片的评分数据;
品质评价模块22,配置为通过深度学习算法结合各图片的评分数据以及特征信息进行训练,得到用于根据图片特征推断图片评分数据的模型;利用训练好的模型对需要评价的地下空间各个地点的多个方位的图片进行评分,并利用各个地点的多个方位的图片的评分数据得到各个地点的评分值;
可视化呈现模块23,配置为按各个地点的评分高低用不同颜色对地下空间平面图上的各个地点进行标注形成评分结果平面分布图并进行显示。
如图3至图5所示,所述图片收集机器人1包括外壳、中央处理单元11、旋转升降装置、拍摄装置13以及移动装置16,所述中央处理单元11位于所述外壳内,所述移动装置16固定于外壳的两端,所述旋转升降装置固定于所述外壳上方,所述拍摄装置13固定于所述旋转升降装置上,所述移动装置16、所述旋转升降装置、所述拍摄装置13均与所述中央处理单元11电连接。图像采集机器人1到达拍摄地点后,中央处理单元11控制移动装置16停止移动,同时启动旋转升降装置和拍摄装置13,每次拍摄时,旋转升降装置升至拍摄装置13距离地面1.70m,模拟人的高度。可选地,所述拍摄装置13包括双摄像头,且两个摄像头之间的间距为人眼间距,模拟人眼看到的空间。所述移动装置16采用履带传动,移动平稳。进一步可选地,所述旋转升降装置包括升降机构12以及固定于所述升降机构12上的旋转机构,所述旋转机构包括相对于升降机构12水平旋转的水平旋转块14以及相对于所述水平旋转块14竖向旋转的竖向旋转块15,所述拍摄装置13与所述竖向旋转块15固定,通过升降机构12实现拍摄装置13的升降,通过水平旋转块14实现拍摄装置13水平方向的旋转,通过竖向旋转块15实现拍摄装置13竖向的旋转,从而可以对地下空间各个地点各个方位的图片进行拍摄,本实施例中主要对每个点的前、后、左、右、上五个方位的 照片进行拍摄。
由于该系统解决技术问题的原理与上述方法实施例相似,故该系统的实施可以参照上述方法的实施,重复之处不再赘述。另外,本申请实施例所提供的几个方法或设备实施例中所揭露的特征,在不冲突的情况下可以任意组合,得到新的方法实施例或设备实施例,在此不进行穷举。
以上所述仅为本申请的可选实施例而已,并不用以限制本申请,凡在本申请的精神和原则之内,所作的任何修改、等同替换、改进等,均应包含在本申请的保护范围之内。

Claims (10)

  1. 一种地下空间品质评价及其可视化呈现方法,包括以下步骤:
    S1、通过图收集机器人收集现有地下空间各个地点的多个方位的图片;
    S2、将收集的图片投入到数据收集平台中,通过众包机制收集网络用户对各图片的空间明暗度、空间舒适性、空间丰富度、环境艺术性四个指标进行评价的结果,并分析得到各图片的评分数据;
    S3、通过深度学习算法结合各图片的评分数据以及特征信息进行训练,得到用于根据图片特征推断图片评分数据的模型;
    S4、重复进行步骤S1~S3,直至模型的准确度达到要求;
    S5、获取需要评价的地下空间各个地点的多个方位的图片以及坐标,利用训练好的模型对各图片进行评分,并利用各个地点的多个方位的图片的评分数据得到各个地点的评分值;
    S6、按各个地点的评分高低用不同颜色对地下空间平面图上的各个地点进行标注形成评分结果平面分布图并进行显示。
  2. 如权利要求1所述的地下空间品质评价及其可视化呈现方法,其中,所述步骤S2包括:
    每次将两张不同地点的相同方位照片或者两张相同地点不同方位的照片投入到数据收集平台中进行显示,利用具有奖励机制的数据收集方式,收集网络用户在预设的关于空间明暗度、空间舒适性、空间丰富度、环境艺术性的问题下,对所给出的两张图片进行对比选择的结果,采用排名算法TrueSkill algorithm根据图片排序得到各图片的评分数据。
  3. 如权利要求1所述的地下空间品质评价及其可视化呈现方法,其中,所述步骤S3包括:
    提取各图片的语义特征和颜色特征,采用全卷积神经网络对所得到的各图片的评分数据以及语义特征和颜色特征进行训练,得到可以根据图片 语义特征和颜色特征推断图片评分数据的模型。
  4. 如权利要求1所述的地下空间品质评价及其可视化呈现方法,其中,所述步骤S5中获取需要评价的地下空间各个地点的多个方位的图片以及坐标包括:如果需要评价的地下空间场地已经建好,则通过图片收集机器人收集该地下空间场地各个地点的多个方位的图片及坐标,如果需要评价的地下空间场地未建好,则根据该地下空间的建筑信息模型BIM进行各个地点的多个方位的截图并记录截图地点的坐标。
  5. 如权利要求4所述的地下空间品质评价及其可视化呈现方法,其中,所述通过图片收集机器人收集该地下空间场地各个地点的多个方位的图片及坐标的方法包括:
    图片收集机器人通过获取地下空间中分布的至少两个基站的信号,判断各个基站的信号角度,通过到达角度AOA算法确定机器人目前相对于基站的坐标,结合基站的地图绝对坐标和图片收集机器人相对于基站的坐标进行计算,获得机器人当前所在位置的绝对坐标,即为该地下空间场地对应地点的坐标。
  6. 一种地下空间品质评价及其可视化呈现系统,用于实现如权利要求1-5任一所述的方法,包括图片收集机器人以及地下空间智能感知平台;
    所述图片收集机器人用于收集现有地下空间各个地点的多个方位的图片;
    所述地下空间智能感知平台包括:
    数据收集模块,配置为将收集的图片投入到数据收集平台中,通过众包机制收集网络用户对各图片的空间明暗度、空间舒适性、空间丰富度、环境艺术性四个指标进行评价的结果,并分析得到各图片的评分数据;
    品质评价模块,配置为通过深度学习算法结合各图片的评分数据以及特征信息进行训练,得到用于根据图片特征推断图片评分数据的模型;利 用训练好的模型对需要评价的地下空间各个地点的多个方位的图片进行评分,并利用各个地点的多个方位的图片的评分数据得到各个地点的评分值;
    可视化呈现模块,配置为按各个地点的评分高低用不同颜色对地下空间平面图上的各个地点进行标注形成评分结果平面分布图并进行显示。
  7. 如权利要求6所述的地下空间品质评价及其可视化呈现系统,其中,所述图片收集机器人包括外壳、中央处理单元、旋转升降装置、拍摄装置以及移动装置,所述中央处理单元位于所述外壳内,所述移动装置固定于外壳的两端,所述旋转升降装置固定于所述外壳上方,所述拍摄装置固定于所述旋转升降装置上,所述移动装置、所述旋转升降装置、所述拍摄装置均与所述中央处理单元电连接。
  8. 如权利要求7所述的地下空间品质评价及其可视化呈现系统,其中,所述拍摄装置包括双摄像头,且两个摄像头之间的间距为人眼间距。
  9. 如权利要求7所述的地下空间品质评价及其可视化呈现系统,其中,所述移动装置采用履带传动。
  10. 如权利要求7所述的地下空间品质评价及其可视化呈现系统,其中,所述旋转升降装置包括升降机构以及固定于所述升降机构上的旋转机构,所述旋转机构包括相对于升降机构水平旋转的水平旋转块以及相对于所述水平旋转块竖向旋转的竖向旋转块,所述拍摄装置与所述竖向旋转块固定。
PCT/CN2021/088812 2020-05-09 2021-04-21 地下空间品质评价及其可视化呈现方法及系统 WO2021227805A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010386869.X 2020-05-09
CN202010386869.XA CN111768361B (zh) 2020-05-09 2020-05-09 地下空间品质评价及其可视化呈现方法及系统

Publications (1)

Publication Number Publication Date
WO2021227805A1 true WO2021227805A1 (zh) 2021-11-18

Family

ID=72719097

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/088812 WO2021227805A1 (zh) 2020-05-09 2021-04-21 地下空间品质评价及其可视化呈现方法及系统

Country Status (2)

Country Link
CN (1) CN111768361B (zh)
WO (1) WO2021227805A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111768361B (zh) * 2020-05-09 2022-12-02 中铁第四勘察设计院集团有限公司 地下空间品质评价及其可视化呈现方法及系统
CN112417556A (zh) * 2020-11-18 2021-02-26 同济大学 基于图片可测度智能评价的bim正向设计方法
CN112905257A (zh) * 2021-01-15 2021-06-04 珠海新势力创建筑设计有限公司 一种基于bim模型自动检测地下室风井及风机房生成条件的软件插件

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105147198A (zh) * 2015-08-10 2015-12-16 深圳先进技术研究院 一种基于扫地机器人的室内地图测绘系统及方法
CN107392252A (zh) * 2017-07-26 2017-11-24 上海城诗信息科技有限公司 计算机深度学习图像特征并量化感知度的方法
CN110926479A (zh) * 2019-12-20 2020-03-27 杜明利 自动生成室内三维导航地图模型的方法和系统
CN111126864A (zh) * 2019-12-26 2020-05-08 中国地质大学(武汉) 一种基于人机对抗评分的街道品质评估方法
CN111768361A (zh) * 2020-05-09 2020-10-13 中铁第四勘察设计院集团有限公司 地下空间品质评价及其可视化呈现方法及系统

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106598081B (zh) * 2016-10-27 2019-08-23 纳恩博(北京)科技有限公司 一种图像采集方法及电子设备
CN111093145A (zh) * 2019-11-15 2020-05-01 垒途智能教科技术研究院江苏有限公司 一种割草机器人的定位系统及定位方法
CN111080167A (zh) * 2019-12-30 2020-04-28 南京慧龙城市规划设计有限公司 一种用于城市规划的地下空间资源质量的评估方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105147198A (zh) * 2015-08-10 2015-12-16 深圳先进技术研究院 一种基于扫地机器人的室内地图测绘系统及方法
CN107392252A (zh) * 2017-07-26 2017-11-24 上海城诗信息科技有限公司 计算机深度学习图像特征并量化感知度的方法
CN110926479A (zh) * 2019-12-20 2020-03-27 杜明利 自动生成室内三维导航地图模型的方法和系统
CN111126864A (zh) * 2019-12-26 2020-05-08 中国地质大学(武汉) 一种基于人机对抗评分的街道品质评估方法
CN111768361A (zh) * 2020-05-09 2020-10-13 中铁第四勘察设计院集团有限公司 地下空间品质评价及其可视化呈现方法及系统

Also Published As

Publication number Publication date
CN111768361A (zh) 2020-10-13
CN111768361B (zh) 2022-12-02

Similar Documents

Publication Publication Date Title
WO2021227805A1 (zh) 地下空间品质评价及其可视化呈现方法及系统
CN107224261B (zh) 利用虚拟现实的视觉障碍检测系统
CN106710360B (zh) 基于增强虚拟现实人机交互的智能驾驶训练系统及方法
Conroy Spatial navigation in immersive virtual environments
CN206431875U (zh) 基于增强现实的医学解剖辅助教学系统
CN105788390A (zh) 基于增强现实的医学解剖辅助教学系统
Thomson et al. Child development and the aims of road safety education
CN107622318A (zh) 一种用于驾考学员预约培训的运营平台及运营方法
CN110969905A (zh) 混合现实的远程教学互动、教具互动系统及其互动方法
CN103877721A (zh) 一种台球辅助训练系统及方法
CN109126056A (zh) 体育训练系统
DE102019102518A1 (de) Validieren von Gestenerkennungsfähigkeiten von automatisierten Systemen
CN109003139A (zh) 一种支持智能交互的文化广场系统及方法
Hsia et al. AI-facilitated reflective practice in physical education: an auto-assessment and feedback approach
CN116611969B (zh) 传统武术智能学习与评分系统
CN110197526A (zh) 一种基于空间无线定位和动作捕捉的应急逃生模拟方法
CN108040239B (zh) 基于图像识别的知识训练系统及方法
CN110909648B (zh) 使用神经网络在边缘计算设备上实现的人流量监控方法
Settgast et al. Hands-free navigation in immersive environments for the evaluation of the effectiveness of indoor navigation systems
CN106814850A (zh) 基于视线轨迹的模拟飞行操作测试系统及测试方法
CN105931509B (zh) 一种电子通信物理智能教学系统
KR101857386B1 (ko) 자기 주도 학습 유도 방법 및 장치
CN114066695B (zh) 一种基于劳动项目的教育内容分析方法、系统和存储介质
VanVoorst et al. Fusion of LIDAR and video cameras to augment medical training and assessment
Inoue et al. Use of human geographic recognition to reduce GPS error in mobile mapmaking learning

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21804343

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21804343

Country of ref document: EP

Kind code of ref document: A1