卷积神经网络用于航空图像上的道路表面分类
Ondrej Pesek1, Lina Krisztian2, Martin Landa1
1Department of Geomatics, Faculty of Civil Engineering, Czech Technical University in Prague, Prague, Czech Republic.
PeerJ. Computer science
|February 3, 2025
概括
卷积神经网络 (CNN) 可以从空中图像中自动分类道路表面. U-Net实现了近92%的准确性,通过展示道路表面细分的优越背景意识,优于随机森林.
科学领域:
- 遥感 遥感 遥感 遥感
- 计算机视觉 计算机视觉
- 地理空间分析是什么
背景情况:
- 道路是人类无处不在的修改,对交通和社会功能至关重要.
- 道路表面类型显著影响其功能,但远程传感数据的自动分类尚未得到充分探索.
- 了解道路表面的组成对于基础设施管理和城市规划至关重要.
研究的目的:
- 调查卷积神经网络 (CNN) 用于使用空中图像进行自动路面分类的有效性.
- 为了比较不同的CNN架构 (FCN,U-Net,SegNet,DeepLabv3+) 的性能,用于道路表面的语义细分.
- 评估近红外波段和过拟合策略对分类准确性的影响.
主要方法:
- 使用10厘米空间分辨率的空中图像进行路面分类.
- 应用并比较了几种CNN模型:完全卷积网络 (FCN),U-Net,SegNet和DeepLabv3+.
- 评估了添加近红外波段的影响,并采用过拟合策略,如放弃和数据增强.
- 将CNN的表现与基于单个像素的随机森林进行比较.
主要成果:
- CNNs成功地区分了紧 (青,混凝土) 和模块化 (铺路石,) 的道路和人行道表面.
- U-Net成为了表现最好的模型,总体准确率接近92%.
- 由于其上下文感知特征提取,CNN在随机森林上表现出显著的优势,U-Net实现了~25%更高的准确性.
结论:
- 卷积神经网络,特别是U-Net,非常适合从空中图像中自动化对道路表面的语义细分.
- 与传统的单像素方法相比,CNN的上下文感知性提供了更高的性能.
- 需要仔细考虑近红外带的纳入和过拟合策略,以获得道路表面分类的最佳结果.
相关概念视频
01:20Deconvolution
129
Deconvolution, also known as inverse filtering, is the process of extracting the impulse response from known input and output signals. This technique is vital in scenarios where the system's characteristics are unknown, and they must be inferred from the observable signals.
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
129
01:24Convolution: Math, Graphics, and Discrete Signals
223
In any LTI (Linear Time-Invariant) system, the convolution of two signals is denoted using a convolution operator, assuming all initial conditions are zero. The convolution integral can be divided into two parts: the zero-input or natural response and the zero-state or forced response, with t0 indicating the initial time.
To simplify the convolution integral, it is assumed that both the input signal and impulse response are zero for negative time values. The graphical convolution process...
To simplify the convolution integral, it is assumed that both the input signal and impulse response are zero for negative time values. The graphical convolution process...
223
01:29Topographic Surveying and Contours
48
Topographic surveying is critical for documenting the Earth's surface, focusing on capturing elevations, slopes, and natural and man-made features. It is essential in construction planning, water resource management, and land-use analysis. The primary outcome of such surveys is a topographic map, which uses contour lines to visually represent the shape and slope of the terrain, providing valuable insights into the landscape's characteristics.Contour lines are fundamental to understanding the...
48
01:17Convolution Properties II
166
The important convolution properties include width, area, differentiation, and integration properties.
The width property indicates that if the durations of input signals are T1 and T2, then the width of the output response equals the sum of both durations, irrespective of the shapes of the two functions. For instance, convolving two rectangular pulses with durations of 2 seconds and 1 second results in a function with a width of 3 seconds.
The area property asserts that the area under the...
The width property indicates that if the durations of input signals are T1 and T2, then the width of the output response equals the sum of both durations, irrespective of the shapes of the two functions. For instance, convolving two rectangular pulses with durations of 2 seconds and 1 second results in a function with a width of 3 seconds.
The area property asserts that the area under the...
166
01:17Design Example: Alignment of a Road Line Using GIS
38
The alignment of a road line using Geographic Information Systems (GIS) is a critical process in civil engineering, combining advanced technology with practical decision-making. This methodology begins with the collection of geospatial data, including information on land cover, geomorphology, drainage patterns, slope, and contour details. Such data is typically acquired through satellite imagery and GIS tools, offering a comprehensive understanding of the terrain.Once the data is gathered, it...
38
01:20Convolution Properties I
134
Convolution computations can be simplified by utilizing their inherent properties.
The commutative property reveals that the input and the impulse response of an LTI (Linear Time-Invariant) system can be interchanged without affecting the output:
The commutative property reveals that the input and the impulse response of an LTI (Linear Time-Invariant) system can be interchanged without affecting the output:
134