Wavelet Transform Based Morphological Matching Area Selection for Underwater Gravity Gradient-Aided Navigation

Selection of gravity gradient matching area is one of the key techniques for underwater gravity gradient-aided navigation. The existing matching area selection methods ignore the high-resolution characteristics of the gravity gradient, resulting in inaccurate selection. Therefore, a frequency domain matching area selection method based on the high-resolution characteristics of the gravity gradient is proposed. The high-frequency information of gravity gradient reference map is extracted by wavelet transform, and the gravity gradient wavelet transform model is established. The morphological image texture segmentation method is proposed to extract the densely textured areas from the gravity gradient high-frequency image as the matching areas. Simulation results show that the proposed method can obtain the texture density, texture amplitude and texture direction in the matching area while obtaining the matching area with a matching rate higher than 90%. Compared with the existing methods, the matching areas obtained by the proposed method are more accurate and the calculation burden is reduced to less than 10% of the existing algorithm. Moreover, the more the trajectory is perpendicular to the texture inside the matching area, the higher is the matching rate.

Automated summary

Selection of gravity gradient matching area is crucial for underwater gravity gradient-aided navigation. Existing methods overlook the high-resolution characteristics of the gravity gradient, leading to inaccurate selection. Thus, a frequency domain matching area selection method based on the high-resolution characteristics of the gravity gradient is proposed. This method extracts the high-frequency information of the gravity gradient reference map using wavelet transform and establishes a gravity gradient wavelet transform model. The proposed method utilizes morphological image texture segmentation to extract densely textured areas as the matching areas. Simulation results demonstrate that this method achieves a matching rate higher than 90% by obtaining texture density, amplitude, and direction in the matching area. Compared to existing methods, the proposed method improves the accuracy of the matching areas and reduces the computational burden to less than 10% of the existing algorithm. Furthermore, the matching rate increases when the trajectory is more perpendicular to the texture inside the matching area.