Remorph框架下的城市智能分析：计算机视觉与听觉融合应用

需积分: 10 20 浏览量更新于2024-07-17 收藏 6.39MB PDF 举报

《计算机视觉与听觉在城市分析中的Remorph框架》一书探讨了人工智能，特别是计算机视觉和计算机听觉技术在城市规划和建筑设计领域的应用挑战。该书通过Tehran Urban Innovation Center（TUIC）组织的一系列名为"Remorph"的工作坊，提供了一种概念框架，旨在整合这两种技术于城市研究之中，推动城市智能化的发展。本书的核心知识点在于： 1. **计算机视觉与城市分析**：计算机视觉作为AI的重要分支，能够解析和理解图像数据，为城市空间的分析提供强大的工具。它可以帮助分析建筑布局、交通流量、环境监测等，通过高精度的图像识别和处理，为城市规划决策提供数据支持。 2. **计算机听觉的应用**：与计算机视觉类似，计算机听觉关注声音信号的处理，包括声音识别、声源定位等。在城市环境中，这可能涉及噪音管理、公共安全预警以及对环境声音的分析，以提升城市居民的生活质量。 3. **Remorph框架**：此框架的设计目的是打破学科间的壁垒，将计算机视觉和听觉技术无缝融合到城市研究中。通过工作坊的形式，参与者可以共同探讨如何优化这两种技术在实际城市问题中的解决方案，如智能交通、灾害预警或城市设计美学的改进。 4. **跨学科融合**：书中的研究不仅涵盖了理论层面，还关注实际应用和未来发展趋势。它强调了如何将工程、计算机科学、物理学、经济学和社会科学等多领域的知识与这些技术相结合，创造出更具创新性和实用性的城市解决方案。 5. **系列出版物的背景**："Studies in Systems, Decision, and Control" 是一个国际学术系列，致力于分享最新的研究成果和前沿进展，为决策制定者、工程师和研究人员提供关于系统、决策和控制的全面视角。《计算机视觉和听觉在城市分析中的Remorph框架》作为其中的一员，体现了这一系列在跨学科研究上的高水准。《Computer Vision and Audition in Urban Analysis Using the Remorph Framework》是一本深入探讨如何利用现代AI技术解决城市问题的著作，展示了其在理论与实践中的重要作用，并通过Remorph框架推动了计算机视觉和听觉技术在城市规划领域的创新应用。

List of Figures

Fig. 1.1 Overview of the book ................................. 2

Fig. 2.1 Polar microphone patterns

.............................. 16

Fig. 2.2 Pointwise sound collection fashion in Tehran

............... 18

Fig. 2.3 Line-wise sound collection fashion in Tehran

............... 19

Fig. 2.4 The perception of HAS from frequency and amplitude [1]

..... 20

Fig. 2.5 The basic concepts of pitch and loudness, which are directly

correlated with frequency and amplitude respectively

......... 22

Fig. 2.6 Concept of DTW. a Euclidian matching. b DTW matching.

c DTW in 3D space this ﬁgure is interpreted from [62, 63]

.... 24

Fig. 2.7 Spectogram of a given sound which is magniﬁed between

0.25 and 0.30 s

...................................... 25

Fig. 2.8 24 Filters are uniformly spaced on the Mel scale

............ 28

Fig. 2.9 Citygram of Tehran

................................... 29

Fig. 2.10 Block diagram of the proposed segmentation technique

....... 31

Fig. 2.11 Urban segmentation technique

.......................... 31

Fig. 2.12 Block diagram of the proposed classiﬁcation technique

....... 32

Fig. 2.13 Urban audio classiﬁcation

.............................. 33

Fig. 2.14 Taxonomy of various sounds in urban environment

.......... 34

Fig. 2.15 An example of functional sound map

..................... 34

Fig. 2.16 The developed GUI for audio classiﬁcation

and segmentation

.................................... 35

Fig. 2.17 Enrollment process for audio classiﬁcation

and segmentation

.................................... 36

Fig. 2.18 The output for a given urban sound

...................... 36

Fig. 2.19 Block diagram of the developed melodiﬁcation techni que

..... 37

Fig. 2.20 GUI for urban melody

................................ 38

Fig. 2.21 Urban melody GUI to track and play simultaneously

......... 38

Fig. 3.1 Kmeans clustering method

............................. 48

Fig. 3.2 The overall computer vision approach

.................... 49

Fig. 3.3 The developed technique for green area detection

............ 51

Fig. 3.4 The ofﬁcial boundary (blue line) on the aerial Google Map

.... 53

xxi

Fig. 3.5 ROI of the aerial image map based on the ofﬁcial

city boundary

....................................... 54

Fig. 3.6 Several ROIs of the images inside the city boundary ......... 55

Fig. 3.7 Green color is extracted from a original to b ﬁltered image

by color range method

................................ 55

Fig. 3.8 Kmeans method for green pixel extraction a original image

b two clusters c three cluster

........................... 56

Fig. 3.9 ROI of the aerial image map based on the separated ofﬁcial city

boundary (barcelona)

.................................. 57

Fig. 3.10 The proposed technique for automatic road detection

......... 58

Fig. 3.11 Original map loaded from Google Map

................... 59

Fig. 3.12 Extracted road or networks by using computer

vision techniques

..................................... 60

Fig. 3.13 Hybrid map (left) and satellite map (right) of same

location

............................................ 61

Fig. 3.14 Final result for water detection based on google

maps fusion

......................................... 61

Fig. 3.15 Desert detection process a original image b green detection

c road detection d water detection e aggregate the water, green,

and road detection f post-processing by morphological image

processing technique

.................................. 63

Fig. 3.16 Original image

...................................... 64

Fig. 3.17 Detected network (road) from original image

............... 65

Fig. 3.18 Detected green from original image

...................... 66

Fig. 3.19 Detected water from original image

...................... 67

Fig. 3.20 Aggregated result from all green, water, and network

detections

.......................................... 68

Fig. 3.21 Apply erosion on aggregated image to remove

built-up areas

....................................... 69

Fig. 3.22 Desert detecting by applying color threshold technique

....... 70

Fig. 3.23 Apply morphological image processing on color detection

result to remove small built-up areas

..................... 71

Fig. 3.24 The proposed built-up detection technique

................. 72

Fig. 3.25 Built-up detection by a threshold around 1

................. 73

Fig. 3.26 Built-up detection by a threshold around 5

................. 73

Fig. 3.27 Built-up detection by a threshold around 10

................ 74

Fig. 3.28 Built-up detection by a threshold around 15 ................ 74

Fig. 3.29 Built-up detection by a threshold around 20

................ 75

Fig. 3.30 Built-up detection by a threshold around 40

................ 75

Fig. 3.31 Built-up detection by a threshold around 100

............... 76

Fig. 3.32 Built-up detection by a threshold around 1000

.............. 76

Fig. 3.33 Shadow detecting by threshold amount of 0.5

.............. 77

Fig. 3.34 Shadow detecting by threshold amount of 0.3

.............. 77

Fig. 3.35 Shadow detecting by threshold amount of 0.2

.............. 78

xxii List of Figures

剩余121页未读，继续阅读

wang1062807258

粉丝: 13
资源: 272

Remorph框架下的城市智能分析：计算机视觉与听觉融合应用

Nginx Cookbook

Theatre In Chicago Audition Manager-crx插件

audition破解

Audition 教程

Audition实验

Audition1.5

audition 3.0

Adobe Audition

audition使用

Audition3.0

最新资源