Clover leaves are a major source of protein for grazing animals in ecological farming. Maintaining a high proportion of clover in the meadow is therefore vital to keeping the animals healthy and happy. Grass-clover systems are however unstable and difficult to manage. The Norwegian Crop Research Institute is therefore investigating how the proportion of clover in meadows can be kept stable over time. SINTEF has developed an image analysis program that helps evaluate the results of the different growing schemes. This way the crop scientists can spend time interpreting results instead of doing timeconsuming and boring data registration.

Evaluation of the different growing schemes requires detailed recordings of what takes place in the field under different conditions both in space and time. Previously the proportion of clover in the meadow has been determined by sorting of samples from bulk or by laborious counting of plants on rather small areas. Neither method is adequate for large scale, detailed analysis of the field. We have therefore developed a program that automatically analyses the content of clover in images of the field at a rate of approximately 40 m² per hour.

Images of the meadow are taken with a standard digital colour camera. Each image is analysed and then added up to a complete map of the clover content of the pasture. We use colour and shape features to classify each pixel as soil, clover, weed or grass. The program gives a classification of each pixel in the images in addition to the overall coverage of clover, grass, soil and weeds for each part of the field.
This makes the program an excellent tool for analysing both spatial and temporal dynamics in grass-clover pastures.

For further information please contact Kristin Kaspersen.