抄録
The harvesting operation is hard work for farmers. Workers hope for a reduction of the harvesting operation in agriculture. The purpose of this study is the construction of a useful visual recognition system that is required for automation and robotization of harvesting operations. In this research, eggplants and tomatoes are taken up as an example of agricultural objects. First, they are searched in a scene image using a genetic algorithm. Second, a recognition method for agricultural objects using Neural-Networks is proposed. In these Neural-Networks, the change of luminance information has been learned as special feature of agricultural objects, which show the luminosity distribution of objects composed of curved surfaces. Finally, through experimental studies, it is shown that the new technology proposed here is effective in the recognition of agricultural object.
