Study On Identification And Feature Extraction For The Typical Pasture Based On Color And Shape

To identify the pasture species is the premise for researching the utilization and evaluation of grassland resource. However, this was fulfilled traditionally by manpower, it is not only a hard working, but also subjective. With the development of machine vision, it has a tendency in realizing the grassland digitalization by recognizing the pasture based on image processing.In this thesis, images of three typical pasture’s leaves (phlomis umbrosa, leymus chinensis and potentilla anserine), taken from the grassland of Wangfu in Siziwang Banner in the Inner Mongolia, were preprocessed first, twelve kinds color moment features and four kinds shape features were extracted, and then BP neural network was built for realizing the image recognition of these three pastures in MATLAB, the image recognition accuracy was 95.6%. The experimental results proved the feasibility of the image identification by using BP neural network based on color moment features and shape features. Compared to the single color feature extraction and recognition, by this method the identification result is more accurate, it has laid a foundation for realizing the digitalization of grassland resources.