Surveillance Bit Rate Control via Perceptual Distortion Tolerance

The patent applies local quality by differentiating bit rate allocation across different image regions based on their perceptual importance. Feature regions (containing objects of interest) are identified and assigned different distortion tolerance levels compared to non-feature regions. This allows the encoder to allocate more bits to perceptually important areas while using fewer bits for less important areas, thereby improving both encoding efficiency and subjective representation quality simultaneously.

The patent changes the control parameter from objective texture complexity to subjective distortion tolerance. By pre-storing correspondence relations between distortion degrees and quality factors, and calculating distortion tolerance for feature regions, the system dynamically adjusts encoding parameters based on perceptual characteristics rather than purely objective measures. This parameter transformation enables the encoder to optimize for human perception while maintaining encoding efficiency.

The patent applies local quality by allocating different quality levels to different regions within the high-resolution image. Instead of uniformly encoding all pixels at maximum quality, the system identifies feature regions and allocates higher bit rates only to those areas, while using lower bit rates for background or less important regions. This regional differentiation maintains perceived image sharpness in critical areas while significantly reducing the total bit rate and network bandwidth requirements.

The patent applies partial action by encoding only the most perceptually important regions at high quality rather than the entire high-resolution image. By calculating distortion tolerance for feature regions and applying selective encoding, the system achieves acceptable or excellent subjective representation with fewer bits than would be required for uniform high-quality encoding of all pixels, thus reducing network bandwidth consumption.