Lossless Compression of Large Field-of-View Infrared Video Based on Transform Domain Hybrid Prediction

Large field-of-view (FOV) infrared imaging, widely utilized in applications including target detection and remote sensing, generates massive datasets that pose significant challenges for transmission and storage. To address this issue, we propose an efficient lossless compression method for large FOV infrared video. Our approach employs a hybrid prediction strategy within the transform domain. The video frames are first decomposed into low- and high-frequency components via the discrete wavelet transform. For the low-frequency subbands, an improved low-latency Multi-view High-Efficiency Video Coding (MV-HEVC) encoder is adopted, where the background reference frames are treated as one view to enable more accurate inter-frame prediction. For high-frequency components, pixel-wise clustered edge prediction is applied. Furthermore, the prediction residuals are reduced by optimal direction prediction, according to the principle of minimizing residual energy. Experimental results demonstrate that our method significantly outperforms mainstream video compression techniques. While maintaining compression performance comparable to MV-HEVC, the proposed method exhibits a 19.3-fold improvement in computational efficiency.