Under water target tracking in radar images using deep learning

Abstract

The scientific community has recently focused on underwater newlinecomputer vision due to the development of autonomous underwater vehicles, newlinesensor technologies, and image processing methods. Due to the climate, newlineinadequate lighting, and inherent turbidity of aquatic environments, newlineprocessing underwater images is generally tricky. A comparative study has newlinebeen done in this field, but it is expanding as the use of autonomous newlineunderwater vehicles rises. Underwater vehicles are primarily utilized for newlineundersea operations, mine detection, and research on aquatic creatures. newlineImages from underwater are recorded using optics, sonar, radar, and newlineultrasound. Optical cameras are frequently helpful for underwater species newlineidentification or counting, coral reefs, pipeline monitoring, mining, etc. newlineBecause of the floating plants and animals, underwater images from such newlinecameras have poor contrast, are blurry, and often contain noise. In turn, this newlinemakes it challenging to recognize objects. Light can only reach up to 100 newlinemeters in pure water, whereas it can only travel a few meters or less in murky newlineor coastal waters. Because light is increasingly reduced when it enters the newlinewater, visibility is restricted. Underwater images can occasionally be taken in newlineconditions with such poor lighting that object detection becomes difficult. newlineEven though underwater vehicles frequently emit light, artificial newlinelight is connected to them. Vehicle movement also contributes to some newlineturbidity in the water, which reduces visibility. Hence, basic pre-processing of newlineunderwater images is essential to precise item recognition. The idea has to be newlineenhanced and denoised because of the poor contrast, uneven lighting, newlineblurring, and other issues. Even though there are established techniques for newlineimprovement, such as histogram equalization and contrast stretching, they fall newlineshort because of the uneven lighting and inconsistent contrast newline