LUCI - Image Segmentation

The original algorithm for this process was developed by Scott Teibert, Doug Aspinall and Mike Duncan (2010) in an effort to create a system capable of isolating the features visible on photographs of farm fields.  The work is predicated on Doug Aspinall's work at OMAFRA on soils and land classification.  This work having been inspired by the work of Bob McMillan and Alex McBratney who are behind the GlobalSoilMap.net project to help farmers in sub-saharan Africa.  The algorithm we have developed differs from previous approaches and has great fidelity in being able to isolate colors that correspond to surface features.  The premise underlying the work is that active hydrological processes tend to collect and focus organic material in a farm field and this results in the patterns of color variation seen in aerial and statellite shots.  One of the more important results that the application of the algorithm reveals is that variability from field to field means that each field tends to have a unique spectrum of colors.  This is true of crops under various forms of stress as well.

The shot above shows a farm in preparation for planting.  The green fields are winter wheat.  Application of the color segmentation algorithm shows the following;

The field on the right of the image clearly shows the banding of colors into three obvious regimes.  Application of an edge detection algorithm gives the following segmentation of the image.

The black boundaries surround the areas of clear color separation.

This separation forms the basis for distinguishing the performance of soils in helping crops grow.  A body of evidence is being formed that shows that these areas correspond strongly to crop yields.  Unfortunately the relationship is not as clear as the color separation, but this forms the first step.

The lighter areas correspond to higher altitudes in the farm field.  Its possible that these higher points can act as the separation points for heated surface air bubbling up into the boundary layer.  If this is the case then its possible that the structure and positions of 'jets' or 'eddies' within the boundary layer are not arbitrary but connected to the topology of the ground.  If the surface topology is having this kind of concentrating and collecting effect on rainfall and irrigation water, then it is also highly likely that the fluid mechanics of heated surface air are similarly affected.