Detection of salient region of in-field rapeseed plant images based-on visual attention model

Detection of salient region of in-field plant based-on visual attention shift with human visual perception was proposed in this study. Based on in-field condition, visual features such as color, intensity and gradient of images, were extracted to build up saliency model with human semantic behavior. Salient regions were obtained by visual attention model and region growing approach. Through analyzing the results from image segmentation experiments, the proposed algorithm has advantage over focusing on rapeseed plant region and avoiding weed regions in outdoor operating environment, which generate a novel idea and method of site-specific weed management.