Integrated approach for ocean data remote sensing with extensive ecological and earth system science learning

Remote sensing pictures are more diversified, and individuals are paying greater attention to marine remote sense studies due to the fast advancement of remote sensing technology. However, due to the complexity of remote ocean data and the varied ocean environments, there is a difference in outcomes even if the same object is recognized in a different scenario. This work provides a pixel-level-based remote-sensing imaging approach (PLORSIA) combining global and regional characteristics to acquire more semantic information and improved pixel predictive capacities. The first is to extract texture, color characteristics, and spatial relationships. Secondly, the algorithm builds a multiresolution local method for cross-attention to get the data about the weight, and regional aspects of ocean remotely sensed pictures in multiple directions. Multiresolution local methods span a comprehensive range of algorithms, models, methods, and concepts. Central to the multiresolution method is somehow to express short-range, mid-range, and long-range relationships explicitly. A multiresolution global cross-attention method is now being developed to achieve global features with an approach. Then each subsystem describes the fusing of global characteristics and local characteristics to get additional depth. Multiresolution global cross-attention network to learn multi-layered depictions of vertices via a Convolutional Neural Network (CNN), and then match the short text snippet with the graphical depiction of the images with the attention mechanisms employed over each layer of the CNN. Lastly, distant ocean-sample sensing is presented via the prediction of the image of pixel level. Three available ocean remotely sensed datasets had tested using the method presented in this article. The empirical findings demonstrate that the PLORSIA process can draw characteristics from tiny samples of remotely sensed pictures from the marine. Furthermore, PLORSIA displays higher prediction skills compared with the forecast outcomes of other remote sensing data systems. The experimental findings show that the suggested model enhances the precision ratio of 93%, recall ratio of 93%, F-score ratio of 93.33%, and accuracy ratio of 90% compared to other existing methods.

Automated summary

This study proposes a pixel-level-based remote-sensing imaging approach that combines global and regional characteristics to improve the predictive capabilities of remote sensing images. The experimental results demonstrate that this method outperforms other remote sensing data systems in terms of prediction accuracy.