The essence of hyperspectral image classification is spectral similarity measure,and some researches indicate that combinative similarity measures can broadly express spectral characteristic,but few people provide systematic evaluation on the topic. Since the correlation coefficient method has a better ability to discriminate the whole shape of spectral curve than the spectral angle cosine method,a new method,Euclidean distance-correlation coefficient,was combined with the former in the paper. Taking spectral curve data and hyperspectral Hyperion image,six similarity measures(i. e.,Euclidean distance,spectral angle cosine,spectral information divergence,correlation coefficient,spectral angle cosine-Euclidean distance and Euclidean distance-correlation coefficient) were comprehensively evaluated under a unified testing framework. Moreover,the characters of these similarity measures were compared and contrasted. The result showed that Euclidean distance-correlation coefficient method would be more precise than other five methods in accuracy of classification.

Taking the standard vegetation spectral library data and the hyperspectral Hyperion remote sensing image, five similarity measure methods (i.e., Euclidean distance, spectral information divergence, spectral angle cosine, spectral correlation coefficient and spectral angle cosine-Euclidean distance) are comprehensively evaluated under a unified testing framework. The results indicate that the spectral angle cosine-Euclidean distance method demonstrates the most superior ability to distinguish various land cover types among five methods because it fully utilizes both the spectral amplitude and shape feature in the hyperspectral data. A combination of the spectral amplitude-sensitive method and the shape-sensitive method will effectively improve the identification accuracy of different land cover types. These evaluation results can be used to guide the selection of an optimal similarity measure method for automatic classification with hyperspectral data.

Taking the vegetation phenological metrics derived from the net ecosystem carbon exchange (NEE) data of 72 flux towers in North America as the references, a comprehensive evaluation was conducted on the three typical classes of remote sensing extraction methods (threshold method, moving average method, and function fitting method) for vegetation phenology from the aspects of feasibility and accuracy. The results showed that the local midpoint threshold method had the highest feasibility and accuracy for extracting vegetation phenology, followed by the first derivative method based on fitted Logistic function. The feasibility and accuracy of moving average method were determined by the moving window size. As for the MODJS 16 d composited time-series normalized difference vegetation index (NDVI), the moving average method had preferable performance when the window size was set as 15. The global threshold method performed quite poor in the feasibility and accuracy. Though the values of the phenological metrics extracted by the curvature change rate method based on fitted Logistic function and the corresponding ones derived from NEE data had greater differences, there existed a strong correlation between them, indicating that the vegetation phenological metrics extracted by the curvature change rate method could reflect the real temporal and spatial variations of vegetation phenology.

This paper presents a phenology-preserving filtering method as a significant improvement to the standard changing-weight filter method to reduce noise in NDVI time series. Specifically it introduces two new features: (1) replacing the changing-weight filter with a 3-point Gaussian filter to improve the computing efficiency; and (2) introducing a multi-year average NDVI time series to remove the false local minima points. This phenology-preserving filtering method was tested at 178 test points for 15 land cover types and 6 test regions around the world using the 250 m 16-day MODIS NDVI product. The results were evaluated in comparison with the original changing-weight filter and other three popular filtering methods. The visual and quantitative analyses demonstrate that the phenology-preserving filtering method can effectively reduce noise and preserve the integrity of the time series with a high computing efficiency.

In order to evaluate the clustering accuracy of different distance measure methods for vegetation index time-series data,we make a comprehensive comparison among six distance measure methods(Euclidean distance,spectral information divergence,spectral angle cosine,kernel spectral angle cosine,correlation coefficient and spectral angle cosine-Euclidean distance) based on the MODIS Enhanced Vegetation Index(EVI) time-series data in China by selecting 468 test pixels across 55 vegetation types and a test region.The test results indicate that the correlation coefficient method shows the lowest clustering accuracy.However,the spectral angle cosine-Euclidean distance method which captures both the curve shape and the amplitude features of the vegetation index time-series data shows the highest clustering accuracy among the six methods.Both the Euclidean distance method which is only sensitive to the spectral brightness and the spectral angle cosine method which is only sensitive to the curve shape perform an inferior clustering accuracy not only in distinguishing different land cover types but also in the regional application.Although the kernel spectral angle cosine method does not show high clustering accuracy in the test at the point level,it shows better performance in the regional application.The spectral information divergence method has a modest performance in the test both at the point level and at the regional level.

This paper evaluated the extraction rate and accuracy of 5 phenology extracting method, taking the vegetation phenological metrics derived from the net ecosystem carbon exchange (NEE) data of 72 flux towers in North America as the reference data. The results indicated that the local midpoint method achieved the highest extraction rate and accuracy. Better performance was observed for the moving average method and the polynomial function fitting method. However, the extraction rate and accuracy of the moving average method were sensitive to the moving window size. The global threshold method performed quite poor. The phenological metrics extracted with the piecewise Logistic function fitting method had a large systematic discrepancy with the NEE derived phenological metrics but there was a strong correlation between them, which indicated that the vegetation phenological metrics extracted with the piecewise Logistic function fitting method can reflect the temporal and spatial variation of vegetation phenology.

Time-series data of normalized difference vegetation index (NDVI), derived from satellite sensors, can be used to support land-cover change detection and phenological interpretations, but further analysis and applications are hindered by residual noise in the data. As an alternative to a number of existing algorithms developed to compensate for such noise, we develop a simple but computationally efficient method (which we call the changing-weight filter method) to reconstruct a high-quality NDVI time series. The new algorithm consists of two major procedures: (1) detecting the local maximum/minimum points in a growth cycle along an NDVI temporal profile based on a mathematical morphology algorithm and a rule-based decision process and (2) filtering an NDVI time series with a three-point changing-weight filter. This method is tested at 470 test points for 55 vegetation types and a test region in China using a 250-m 16-day Moderate Resolution Imaging Spectroradiometer (MODIS) NDVI product. Comparing our results to those of three other well-known methods-asymmetric Gaussian function fitting, double logistic function fitting, and Savitzky-Golay filtering-the new method has many of the advantages of existing methods, while in some cases, the changing-weight filter method more effectively preserves the curve shape as well as the timing and the amplitude of the local maxima/minima in the NDVI time series for a broad range of phenologies. Moreover, the response of the filtering algorithm is relatively insensitive to the exact values of its design parameters, making the new method more flexible and effective in adjusting to fit a variety of classes of NDVI time series.

Extracting cropland parcels from high-resolution remote sensing images is an important issue for dynamic land-use monitoring, precision agriculture and other fields. However, cropland spectra change frequently in time and spatial space. The application of multi-spectral image classification in cropland extraction, not only leads to misclassification with other vegetation easily, but also results in broken parcels caused by salt and pepper effect. Texture is an important feature of satellite images, which takes into account pixel gray scale difference and the spatial relationship between neighboring pixels. In order to overcome the impact of spectral variability, this paper presents an advanced cropland parcel extraction method based on texture analysis and multi-spectral image classification. Test on an ALOS (Advanced Land Observation Satellite) image shows that this method can effectively reduce the impact of spectral variations and obtain satisfactory results. But there still has some aspects which should be further improved in the future study, including: (1) some "noise" polygons still exist because the filter can not eliminate all the noise pixels completely; and (2) parcels generated by this approach can not reflect their subtle internal difference, such as inner boundary shaped by different crops.