Using Prior Knowledge to Resolve Complex Spatial Patterns in Agriculture
Complex spatial patterns made up of different land uses, soil changes, disturbance, different image dates and even wetness and burning patterns are difficult to resolve using conventional image analysis techniques. Removing these confusing patterns using field checking to identify the target pattern is laborious, time consuming and expensive. It also negates the main advantage of using remote sensing, that of rapid, synoptic assessment. Applying the principle of signal detection to these spatial patterns can allow the progressive removal of known, or suspected, patterns leaving the spatial pattern of the target variable clear. Application of this approach requires only a few assumptions to be made and it can be done rapidly and cheaply. In a test example of this technique, land use and other cultural patterns are removed from an image leaving the residual pattern of soil geochemistry. Field checking of the result showed this to be a successful and economical approach.
Spatial Data Signal Detection Theory Agriculture Remote Sensing
Brian Lees
School of Physical, Environmental and Mathematical Sciences, UNSW@ADFA, Canberra, ACT 2600, Australia;Institute of Soil and Water Conservation, Chinese Academy of Sciences and Ministry of Water Resources, (ISWC, CAS&MWR)
国际会议
西安
英文
1125-1129
2010-10-11（万方平台首次上网日期，不代表论文的发表时间）