Spatial Filtering for Video Compression Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion compensated interframe coding methods suffer from significant quantization noise in high complexity, high motion scenes, leading to wasted bandwidth, and introduce additional delay that is unacceptable in applications like videoconferencing.
Innovation Solution
Implementing a spatial recursive filter with adaptive coefficients to reduce noise in the difference picture while maintaining a constant delay of one line plus one pixel, allowing for efficient noise reduction without degrading communication synchronization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If temporal recursive filtering is used to reduce quantization noise, then noise reduction effectiveness is improved, but additional delay is introduced that degrades communication synchronization
Solution Approach 1:
The patent transitions from temporal filtering (frame-to-frame processing in the time domain) to spatial filtering (pixel-to-pixel processing within a frame). By applying spatial recursive filtering to the difference picture using horizontal and vertical filter coefficients, the system reduces quantization noise without introducing the additional frame-level delay characteristic of temporal filtering, thus resolving the contradiction between noise reduction effectiveness and communication synchronization.
2Productivity
If spatial filtering is applied to reduce noise in difference picture, then compression efficiency is improved, but processing complexity increases
Solution Approach 1:
The patent divides the spatial filtering operation into separate horizontal and vertical filtering stages, each processing specific pixel relationships within the difference picture. This segmentation allows the complex spatial filtering to be implemented through simpler, standardized filter coefficient applications, improving compression efficiency while managing processing complexity through modular operation.
Solution Approach 2:
The patent uses adaptive filter coefficients that vary based on image content and position. By changing the filter strength parameters dynamically rather than applying a fixed complex filter, the system achieves improved compression efficiency while keeping the processing algorithm relatively simple and manageable.
3Reliability
If de-blocking filter is applied to filter block edges, then visual quality is improved, but processing time is increased
Solution Approach 1:
The patent merges the de-blocking filter operation with the spatial filtering process by integrating the block edge filtering into the broader spatial recursive filtering framework. This combination allows visual quality improvement through unified filtering rather than separate sequential operations, reducing total processing time while maintaining effective block edge filtering.
Data Source
AI summary
An apparatus, method, and carrier medium carrying computer-readable code to implement a method that includes generating a difference picture of an accepted picture of a time sequence of pictures minus an estimate of the previous picture in the sequence. The generating of the difference picture includes forming quantized coefficients and passing the quantized coefficients via a prediction loop to generate the estimate of the previous picture. A spatial filter is applied to at least a component of the difference picture such that the forming of quantized coefficients is from a spatially filtered difference picture. The spatial filter is a noise reducing spatial filter configured such that there is a fixed delay between the filter input picture and the filter output picture. The fixed delay is significantly less than the time between consecutive pictures in the time sequence.