IMPROVING DRONE IMAGE COMPRESSION VIA DEEP LEARNING FOR TARGETED REGION EXTRACTION

Abstract

Image compression is a critical technique for minimizing the storage and transmission requirements of digital images. Traditional methods like JPEG and JPEG2000 have reached their limits in terms of performance improvement after years of development. Recently, there has been a growing interest in leveraging deep learning to develop more efficient image compression methods. Deep learning-based approaches eliminate the need for manual module design and optimization, allowing neural networks to construct and jointly optimize compression modules, resulting in the intelligent removal of image redundancy. This approach holds promise for significantly enhancing image compression performance. This paper explores the potential of deep learning in image compression and its ability to create more efficient compression methods by leveraging neural networks for intelligent redundancy reduction.