*5.2. Method Repeatability: Applicability of the Proposed Method in another Situation*

To further verify the performance of the proposed method in other waters when adopting different kinds of sonars, another experiment was conducted in BeiBu Gulf. In this area, the depth ranges from 40 to 50 m; seabed sampling shows that sediments mainly consist of sand, which has two forms of existence: sand waves and ripples. Klein 3000 and EM710 were separately adopted for SSS and MBES measurement. After processing the obtained measurement data—the seabed topography is shown in Figure 11—the SSS and MBES images were presented in Figure 12. It can be seen that the seabed topography and the MBES image can only reflect the sand waves but the SSS backscatter image can also capture the small scale sand ripples clearly. This difference can be seen in areas (a1) and (b1) of Figure 11. This difference can also be seen in zoomed areas (a1) and (b1) of Figure 12.

Based on the common sand wave features reflected by both images, the proposed image matching method described in Section 2.2 was performed and the matched result shown in Figure 12 proved that the method can also be effective when using other kinds of measurement instruments in a different water area. An area with sharper edges is zoomed in for better visualization, as shown in Figure 12c. Using the correct matches that were finally obtained to rectify the SSS backscatter image geographic coordinates, the obtained SSS backscatter image was superposed on the seabed topography. The comprehensive presentation of 3D topography and surface details is shown in Figure 13. Compared with Figure 11, it can reflect both the obvious sand waves and detailed sand ripples.

**Figure 11.** Seabed topography of the new water area.

**Figure 12.** The matched result of SSS backscatter image (**a**) and MBES image (**b**) of the new water area. The (**a1**) and (**b1**) are the zoomed areas; (**c**) is the example with sharper edges; the colored full lines in (**a**) (**b**) and (**c**) are used to connect matched points.

**Figure 13.** Comprehensive seabed topography and surface details of the new water area.

#### *5.3. Sonar Frequencies*

The acoustic waves emitted by a sonar propagate through sea water to seabed and the backscattered echoes are received by the transducer to form seabed backscatter images. When using different frequencies of acoustic wave, the impacts of surface and volume scatter are different [45–47]. Taking the sandy sediments as an example, penetration depth is limited to 12mm for 500 kHz, while 100 kHz may penetrate 90mm into the subsurface [48]. In addition, the acoustic frequency affects sonar backscatter image texture presentations due to changing surveying parameters, such as footprint sizes and different sensitivity to seafloor features [23,45]. As a result, the presentation of seabed features may be different when using different SSS and MBES frequencies [23,45,47]. As a result, finding matches might become difficult.

At specific frequencies, the sonar backscatter images can always reflect seabed features and the relative positions of these features are stable. Therefore, based on the geometrical structure formed by the association of three detected features, image matching can still be conducted [43]. Meanwhile, deep learning has also been applied in the sonar image processing field [49,50] and some related image matching methods [51] can be used for sonar image matching.

Recently, multi-frequency SSS and MBES have been applied, such as EM2040 and Edgetech 4200SP. As for multi-frequency sonar backscatter data, they can be rendered as acoustically colorful images [23,46]. The formed colorful image can reflect more seabed features compared with those obtained from monochromatic sonar with a single center frequency. Consequently, more features can be detected from the colorful image and can be used for image matching.

#### **6. Conclusions**

Despite some new sonar systems appearing that can provide co-registered seabed images and topography, they can only be used in restricted waters. The SSS and MBES are still the most effective sonars for obtaining seabed images and topography, especially for deep seafloor observation. Based on the complementarity of SSS and MBES data, this paper proposes a new method for acquiring high-resolution seabed topography and surface details by combining SSS and MBES data. Through taking the image geographic coordinates as the constraint when using the SURF algorithm for initial image matching, the authors have obtained more correct initial matched points compared to that without constraint. Then, the finer matching step is conducted by adopting a template matching strategy, which uses the DLSS descriptor to reflect the shape properties of the area centered feature points. The combination of the initial and finer matching steps can help improve the matching performance and the percentage of correct matched points rose to 86%. Based on the obtained matched points and considering the difference of geographic distortions in different parts of the SSS backscatter image, the SSS backscatter image was segmented into several blocks and the geometric transformation model within each block can be established to rectify its geographic coordinates. Subsequently, the rectified SSS backscatter image can be superposed on the topography, which can reflect both the 3D seabed undulations in combination with a high detailed SSS backscatter imagery.

Experiments have verified this method. By this method, the high-resolution and high-accuracy seabed topography and surface details can be represented together, which is meaningful for understanding and interpreting seabed topography. Meanwhile, this paper discusses the accuracy of the reckoned SSS positions and uses it as a reference threshold in the image matching process. In addition, this paper discusses the impact of sonar frequency on the sonar backscatter image and provides some useful suggestions when dealing with multi-frequency sonar image matching.

**Author Contributions:** Conceptualization, X.S. and J.Z.; Data curation, X.S. and J.Z.; Formal analysis, X.S., J.Z. and H.Z.; Funding acquisition, J.Z. and H.Z.; Methodology, X.S. and J.Z.; Supervision, J.Z. and H.Z.; Visualization, X.S., J.Z. and H.Z.; Writing—original draft, X.S.; Writing—review & editing, X.S., J.Z. and H.Z.

**Funding:** This research was funded by the National Natural Science Foundation of China, grant numbers 41576107, 41376109 and 41606114.

**Acknowledgments:** The authors would like to thank the editors and anonymous reviewers for their comments and suggestions. The data used in this study were provided by the Guangzhou Marine Geological Survey Bureau. The authors are grateful for their support.

**Conflicts of Interest:** The authors declare no conflict of interest.
