遥感时序去噪算法的系统评价
Systematic Evaluation of Remote Sensing Time Series Denoising Algorithm

近年来，大批量煤矿开采已经严重破坏到矿区周围的生态环境，而新时代我国注重绿色发展，这就需要对矿区生态条件进行修复。修复生态可从分析研究该区域的植被覆盖度开始。NDVI时间序列数据是目前最常用的数据源之一，其在全球气候环境变化、植被覆盖动态变化检测、植被信息提取等方面具有广泛应用。但由于传感器、云层覆盖等因素影响，NDVI时序数据比一般数据更容易出现噪声，因此时序数据去噪显得至关重要。文章以宝日希勒矿区为研究区，从目视判别时序曲线、对比去噪样本点的去噪均方根误差、以及变化检测结果精度分析三个角度，在matlab编程软件以及ENVI软件的帮助下，实现这三种去噪算法，并对比分析这3种方法的优点和缺点。研究结果表明：1) 3种去噪算法都有效地对原始NDVI时序数据进行了去噪，且效果明显；2) 不同的去噪方法均存在着过度拟合的现象；3) 对于文章所选的NDVI最大合成数据而言，B-W去噪效果最好，其次是去噪效果一般的BISE算法，S-G算法的去噪效果最差。
In recent years, large-scale coal mining has seriously damaged the ecological environment around the mining area. In the new era, China pays attention to green development, which requires the restoration of the ecological conditions of the mining area. Ecological restoration can start from analyzing the vegetation coverage of the area. NDVI time series data is one of the most commonly used data sources at present. It has been widely used in global climate and environmental change, dynamic change detection of vegetation cover, vegetation information extraction and so on. However, due to the influence of sensors, cloud cover and other factors, NDVI time series data are prone to noise, so data denoising is very important. Taking Baorixile mining area as the research area, this paper realizes these three denoising algorithms under Matlab programming software from three angles: visual discrimination of time series curve, comparison of denoising root mean square error of denoising sample points, and accuracy analysis of change detection results, and compares and analyzes the advantages and disadvantages of these three methods. The results show that: 1) The three smoothing algorithms all effectively denoise the original NDVI time series data, and the effect is obvious; 2) Ecological restoration can start from analyzing the vegetation coverage of the area; 3) For the NDVI maximum synthetic data selected in this paper, B-W has the best denoising effect, followed by bise algorithm, and finally S-G algorithm has the worst denoising effect.