A Novel Estimation Method of Grassland Fractional Vegetation Cover Based on Multi-Sensor Data Fusion

Yuzhuo Zhang,Tianyi Wang,Yong You,Decheng Wang,Mengyuan Lu,Hengyuan Wang

DOI: https://doi.org/10.1016/j.compag.2024.109310

IF: 8.3

2024-01-01

Computers and Electronics in Agriculture

Abstract:Grassland Fractional Vegetation Cover (FVC) needs to be monitored because it can reflect the ecological situation of grasslands. Currently, there is a lack of efficient detection methods in near-earth remote sensing detection systems. Traditional image methods have the problem of detecting redundant vegetation information under high coverage and detecting missing vegetation information under low coverage. This study proposed a novel estimation method based on multi-sensor data fusion (RGB, thermal infrared spectrum and depth point cloud information fusion). This fusion method takes advantage of each type of data to estimate different FVC ranges. Regarding RGB image segmentation, this study improved network module based on BlendMask and proposed a Transformer-based GANet-Mask algorithm that can effectively detect vegetation information within the FVC range of 30% to 70%. GANet-Mask was more than 5% lower than BlendMask in actual detection coverage loss rate (RLOSS) and was better than other instance segmentation models. In the range of FVC below 30% and above 70%, this study experimentally proved the effectiveness of the two fusion methods. Experiments have demonstrated that the fusion method can reduce the RLOSS estimated by a single RGB by more than 5% (even the GANet-Mask model) and reduce the RLOSS by more than 10% compared with the non-fusion method. In the final ten sets of field random experiments, the average precision of the 10 test sites was 94.3%, the average recall was 93.5%, and the average F1 was 93.8%. At the same time, this study conducted ten sets of field experiments with different FVCs. The average accuracy rate of the ten groups was 94.1%, which can fully meet the requirements of grassland vegetation. This study may contribute to grassland intelligence monitoring and grassland protection.