一种基于AS-Unet++的语义细分方法,用于对图像进行功率遥感
Guojun Nan1, Haorui Li1, Haibo Du1
1School of Electrical Engineering and Automation, Hefei University of Technology, Hefei 230009, China.
Sensors (Basel, Switzerland)
|January 11, 2024
概括
本研究介绍了AS-Unet++,一种用于增强远程传感图像分析的新语义细分方法. AS-Unet++显著提高了用于电力输电线路规划的特征识别精度.
科学领域:
- 计算机视觉 计算机视觉
- 遥感 遥感 遥感 遥感
- 人工智能的人工智能
背景情况:
- 电力传输线路的自动规划需要精确的特征识别从遥感图像.
- 遥感数据中的特征信息显示出不同的深度和不均的分布.
- 现有的方法可能无法充分捕捉复杂地形的多深度特征.
研究的目的:
- 提出一种新的语义细分方法,AS-Unet++,用于改进远程传感图像中的特征识别.
- 提高在不同深度捕捉特征的能力,提高识别准确度.
- 为了减少训练参数的数量,同时保持或提高性能.
主要方法:
- 通过将Atrous空间金字塔聚合 (ASPP) 和Squeeze-and-Excitation (SE) 模块集成到传统的U-Net架构中开发了AS-Unet.
- 通过堆叠AS-Unet的不同层来构建AS-Unet++以使多深度特征的学习成为可能.
- 实施了修剪机制,以优化网络深度,并在确定最佳配置后减少参数.
主要成果:
- 与标准的U-Net.相比,AS-Unet++在不同深度的自动学习功能方面表现出卓越的性能.
- 拟议的方法实现了远程传感图像特征的识别精度的显著整体改进.
- 修剪策略有效地减少了训练参数的数量,从而产生了更高效的模型.
结论:
- AS-Unet++为远程传感图像分析中的语义细分提供了强大而准确的解决方案.
- 该架构有效地解决了因不同特征深度和不均分布所带来的挑战.
- 这一进步有助于通过增强的图像识别来更精确地自动规划输电线路.
相关概念视频
Deconvolution
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...
Upsampling
Managing signal sampling rates is essential in digital signal processing to maintain signal integrity. A decimated signal, characterized by a reduced frequency range due to its lower sampling rate, can be upsampled by inserting zeros between each sample. This upsampling process expands the original spectrum and introduces repeated spectral replicas at intervals dictated by the new Nyquist frequency. To refine this zero-inserted sequence, it is passed through a lowpass filter with a cutoff...


