Gravitational Texture Feature Based Directional Adaptability Analysis Method for Gravity-Aided Navigation

Gravity aided inertial navigation is the key technology for underwater vehicle, the selection of matching area has a significant impact on navigation accuracy. The fluctuation of gravity anomaly in the gravity background field exhibits irregular characteristics, and the navigation accuracy is different when underwater vehicle enters in different directions. In order to evaluate the directional adaptability of the gravity filed, a directional adaptability analysis method based on gravitational texture features is proposed. To achieve directional evaluation of the matching area, the method extracts the gravity anomaly gray-gradient co-occurrence matrix and adopt multidirectional Sobel operators to perform gradient operations in different directions. In addition, a parallel convolutional neural network model is developed based on statistical feature parameters and gravitational texture feature parameters, propose the concept of direction matching rate, effectively achieve objective evaluation of the gravity field and avoid the influence of subjective factors on the selection results. Simulation results show that compared with the traditional method, the proposed method effectively select the matching area with rich gravity information and improves the multidirectional matching rate for gravity aided navigation.