基于地质统计学影像纹理的石河子地区化控期棉花长势监测

Cotton growth monitoring during chemical control stage using geostatistical image texture: a case study of Shihezi

化学调控是棉花生长期间的一个重要环节,在较大区域尺度上,化控常常整齐划一,依固定的量而行,忽视其普遍存在的空间差异,结果导致某些棉田投入过量或投入不足,在精细农业所倡导的基于特定场的农业投入这一框架下,为辅助决策者制定有针对性的棉田化控方案,该文提出了应用地质统计学影像纹理进行对化控棉花长势监测,通过借助地质统计学方法在分析空间结构变化上的优势,强化遥感影像上不同长势棉田间的差别,提高样本选择的分离度和长势空间分类精度。实验结果表明,运用变异函数纹理结合光谱波段的最大似然分类方法能够很好地界定棉花长势的不同类型,总分类精度达90.53%,证明了应用地质统计学影像纹理方法在针对化控的棉花长势监测上的有效性。

The chemical control is an important input during the course of the growth of cotton, for a great scale region, the date of chemical control usually refers to some solid season, regardless of the spatial difference among cotton fields. Under the framework of precision farming, this paper recommends using geostatistical image texture to monitor the growth cndition of cotton, guiding the spray of chemical medicine. Comparing with tradition method based on vegetation index, although it could not calculate the concrete value of cotton growth condition, it could precisely abtain the spatial distribution of growth condition, what is more, it could identify the cotton field which is at the vigorous growth stage, and which is not able to be distinguished by the method based on vegetation index, so it is a better method towards guiding to chemical control of cotton.