Feature Map Compression for Video Coding for Machines Based on Receptive Block Based Principal Component Analysis.
Minhun LeeHansol ChoiJihoon KimJihoon DoHyoungjin KwonSe-Yoon JeongDonggyu SimSeoung-Jun OhPublished in: IEEE Access (2023)
Keyphrases
- video coding
- feature maps
- motion compensation
- principal component analysis
- motion estimation
- image and video compression
- compression efficiency
- motion estimation and compensation
- bit allocation
- motion compensated
- bit rate
- rate distortion
- video compression
- motion vectors
- feature space
- video signals
- block size
- temporal redundancy
- self organizing maps
- macroblock
- efficient compression
- kernel function
- video quality
- video codec
- dimensionality reduction
- video coding standard
- object based coding
- computational complexity
- discrete cosine transform
- compression algorithm
- saliency map
- face recognition
- bitstream
- feature vectors
- optical flow
- neural network
- video sequences
- image compression
- face images
- feature extraction
- video coder
- coding efficiency
- video data
- inter frame
- support vector machine
- low bit rate
- natural images
- distributed video coding
- subband
- multiscale
- visual quality
- compression ratio
- real time