Wheat Density Estimation Method Based on Multi-Sensor Information Fusion

By using multi-sensor information resources for multi-information fusion, a large amount of information describing crop characteristics can be obtained. Compared with single detection technology, multi-information fusion technology has the advantages of rich information and good fault tolerance. In this paper, we exploit the complementary strengths of the two sensors by using a sensor fusion approach to explore 3D object detection methods for crops in dense scenes. This method fuses the images and point cloud data collected synchronously by the binocular camera and lidar, uses visual detection to generate a two-dimensional bounding box for each object, and projects and matches these two-dimensional bounding boxes with the point cloud to obtain the corresponding three-dimensional point cloud. In order to avoid the multi-ear situation existing in the 2D detection frame, we adopted a clustering method based on the 3D detection frame. Finally, we output the detected object information in the form of point cloud and 3D bounding box, so as to realize the detection of wheat density. This method can be applied to the complex environment where wheat ears grow densely, and can greatly improve the accuracy and robustness of wheat ear location.