Navigation Path Points Extraction Method Based on Color Space and Depth Information for Combine Harvester

Automatic harvesting is an important part of agricultural automation. Navigation path points extraction is a prerequisite of automatic harvesting for combine harvester. The traditional methods of navigation path extraction in the farmland are mainly based on color space. The color space can be heavily influenced by environmental factors, such as illumination, weeds, shadows and crop types. On the contrary, depth information can be less affected by these factors. This paper proposes one method about navigation path points extraction based on both color space and depth information, which can improve robustness and accuracy of the navigation path points extraction. The method is mainly composed of three parts: navigation path points extraction based on color space, navigation path points extraction based on depth information and navigation path points fusion. The farmland experiments show that an average accuracy of navigation path points extraction is 98.02% with average pixel error of 4.08, average relative error of 0.64% and average distance error of 3.3cm. The experiment results show that this paper proposes one effective navigation path points method.