Browsing by Subject "Juuri"

Nowadays there is a lot of discussion about climate change and climate change affects everyone. In agriculture when studying the carbon cycle, it is essential to know and understand what is happening underground. Organic carbon bound to the soil is about 1500 Gt when measured from a depth of 1 m. According to the datasets in topsoil, at a depth of 0–30 cm, the amount of organic carbon is about 50 % of the total amount of organic carbon. For surface and subsoil examination the minirhizotron system provides an inexpensive and simple way to examine the roots of plants without destroying the plants examined. The objective of this study is to build and test a minirhizotron camera and scanner as well as to become familiar with image processing and its automation. The construction of the camera was started from an old minirhizotron camera which could be used as parts for a new one. The parts which could not be used from the old camera were designed with 3D modeling software and manufactured with 3D printer. For image acquisition an electronic microscope was bought. For the scanner system a normal USB powered flatbed scanner was purchased and a protective case was built for it. Imaging with a minirhizotron camera transparent tubes must be installed to the ground and they are usually installed vertically, horizontally, or at an angle of 30 ° to 45 °. The way that a scanner is installed is usually determined by what the purpose of the imaging is. The obtained image material showed that the self-built minirhizotron system is a suitable tool for root research and its relatively inexpensive price allows a wide distribution of the system, even among ordinary farmers. A prerequisite for the success of the images is close contact between the ground and the tubes or scanner. It allows the images to be accurately thresholded and gives reliable results when analyzing the images. Highlighting the roots by hand is very time consuming which is the reason that development of automatic analyzing is needed. Successful thresholding is also a prerequisite for the automation and it should be noted in the future. Viability of machine vision and machine learning in root studies should be studied.