The item attributes concerned in the classification have been normally labeled as personalized, spectral, spatial and texture. 1290543-63-3Fig 5 only shows the 28 attributes with relative significance increased than one in the prediction product. Most of these characteristics belonged to the spectral category, adopted by the customized , texture and spatial attribute categories. We speculated that the highly fragmented landscape in the study location resulted in the extremely diversified geometric styles, even for the exact same crops. The results of the texture attributes on the classification have been not substantial and were possibly triggered by the relative coarse spatial resolution of the HJ-1 CCD picture data.Between the 18 spectral characteristics, six, 5 and five attributes ended up related to the purple, eco-friendly and blue bands, respectively. Only two attributes were associated to the NIR band, and the relative significance of these characteristics was small. This locating may possibly be right related to the low significance of vegetation indices in the prediction product. Additionally, this obtaining indicated that the combined single bands could accomplish seem classification outcomes without having fully employing the vegetation index details.The phenologies of the significant crops in the examine location could supply crucial info for selecting remote sensing photographs. The six images utilized in this examine lined the whole sugarcane increasing season and included the main phenologies of rice and peanut. For instance, the 1st temporal time period was June, when the blue band spectral reflectance could be utilised to obviously discriminate the crops from eucalyptus and banana. In Fig 6, the No. five node, which corresponds to the graphic from October, employed the optimum red gentle absorption capability of sugarcane at that stage to differentiate among rice and peanut. When using information of the phenologies of the significant crops in the particular review region, the impression choice need to be really specific and the prediction design should be effortless to interpret.From Figs five and 6, we observed that the very first four photographs dominated the building processes of the choice policies . As a result, beneath this technological framework, we built understanding regarding which vital temporal window need to be utilised to information graphic choice. Subsequent, the redundant photos of these ranges could be safely omitted. For this particular research, if only many experienced photographs masking the early and middle sugarcane phenologies can be acquired, then the classification accuracy should be certain. Furthermore, the remote sensing photos attained during the latter phenologies soon after sugar accumulation could not be needed.The classification of sugarcane in southern China in big locations faces two issues: a limited quantity of experienced remote sensing knowledge because of to pervasive cloudy weather conditions and the complex combination of land protect and their related spectral reflectances. In this context, our goal was to entirely use the spectral and textural differences in a variety of croplands in limited middle-resolution remote sensing pictures to aid the classification of sugarcane. Moreover, we aimed to determine whether a appropriate temporal window exists to information the variety FK866of important remote sensing photographs in sugarcane classification.In this study, six HJ-one CCD pictures with a spatial resolution of thirty m and masking the total sugarcane increasing time have been utilised.