Estimating grass-clover ratio variations caused by traffic intensities using image analysis

Grass and especially clover have a negative yield response as a function of  traffic intensity.  Conventional grass-clover production for silage have high traffic intensity due to fertilizing with slurry, cutting the grass, rolling the grass into swaths, and collecting and chopping the grass into trailers with a forage harvester.  Normally, the traffic is spread throughout the whole field area during the growth season. Hence, the tracks formed by the machines will have different impacts on the grass and clover growth and yield.   Compared to grass, clover is known to have a higher feed value[1]. In order to evaluate the impact of different traffic intensities on the grass-clover, it is essential to know the ratio between grass and clover.   The objective of this paper was to develop and evaluate an automated image acquisition and image analysis method capable of estimating the grass clover ratio within the context of a full scale field trial. A 14 hectare full scale grass-clover field trial with 24 different traffic intensities and 35 replicates was established. Each net parcel measured 9 x 1.3 m and the 24 treatments were randomized onto the 840 net parcels. The grass clover was established in spring 2007 using RTK-GPS auto steered tractors and implements. A Claas Axion tractor equipped with AutoFarm RTK AutoSteer guidance system was used to carry two parallel mounted cameras over the net parcels at a speed of 1 m s^-1. In combination, the cameras sampled an area of 1.3 x 0.48 m with 2.1 Hz ensuring that the whole parcel was imaged. Each image was geo-positioned. The image analysis comprised two steps: Extraction of green material and discrimination of grass and clover using the morphological opening approach. This paper shows the initial results using the automated imaging analysis algorithm to fulfill the aim

[1]http://www.ruralni.gov.uk/index/publications/press_articles/dairy-2/role-of-clover.htm