Spatial-Temporal Context Model for Remote Sensing Imagery Compression

Jinxiao Zhang, Runmin Dong, Juepeng Zheng, Mengxuan Chen, Lixian Zhang, Yi Zhao, Haohuan Fu

Abstract: With the increasing spatial and temporal resolutions of obtained remote sensing (RS) images, effective compression becomes critical for storage, transmission, and large-scale in-memory processing. Although image compression methods achieve a series of breakthroughs for daily images, a straightforward application of these methods to RS domain underutilizes the properties of the RS images, such as content duplication, homogeneity, and temporal redundancy. This paper proposes a Spatial-Temporal Context model (STCM) for RS image compression, jointly leveraging context from a broader spatial scope and across different temporal images. Specifically, we propose a stacked diagonal masked module to expand the contextual reference scope, which is stackable and maintains its parallel capability. Furthermore, we propose spatial-temporal contextual adaptive coding to enable the entropy estimation to reference context across different temporal RS images at the same geographic location. Experiments show that our method outperforms previous state-of-the-art compression methods on rate-distortion (RD) performance. For downstream tasks validation, our method reduces the bitrate by 52 times for single temporal images in the scene classification task while maintaining accuracy.

Primary Subject Area: [Experience] Multimedia Applications

Secondary Subject Area: [Systems] Transport and Delivery

Relevance To Conference:

This paper introduces a Spatial-Temporal Context model (STCM) designed for remote sensing (RS) image compression, which can contribute to multimedia/multimodal processing in the following aspects:
Firstly, image compression is an important technology and research direction in multimedia applications. It reduces the storage space required for images, facilitating efficient transmission over networks and enabling faster processing in multimedia systems and applications. Efficient image compression techniques have become imperative due to the growing availability of source data from diverse fields, including remote sensing areas.
Secondly, remote sensing images comprise data from different satellite sensors, containing diverse modalities of information. Different sensors capture images with distinct characteristics, including variations in spatial resolution and spectral wavelength and numbers. Based on our model, we propose to adapt different spectral and temporal numbers of remote sensing images during the inference stage, effectively accommodating the varied modalities inherent in RS imagery.

Submission Number: 370