Quantization Control Apparatus for Video Encoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video encoding techniques using intra slices face challenges in maintaining accurate complexity indices when the texture of the encoding target picture differs significantly from the previous picture, leading to degraded image quality, especially in methods like H.264 that rely on intra prediction models.

Innovation Solution

A quantization control apparatus and method that computes similarity between the intra slice region of the encoding target picture and previously-encoded pictures, selects a picture with the highest similarity, and determines the quantization step based on the selected picture's encoding information to improve the accuracy of complexity indices and maintain image quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If the complexity index is computed using encoding information from the immediately preceding picture, then the computation is simple and fast, but the accuracy of the complexity index degrades when the texture of the encoding target picture differs significantly from the previous picture

Engineering Contradiction:
Improveaccuracy of complexity indexVSAvoidcomplexity of picture selection process
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent pre-computes and stores encoding information (code amount and quantization step) for multiple previously-encoded pictures before the encoding target picture is processed. This preliminary preparation allows the system to quickly retrieve and compare encoding information from multiple candidate pictures without increasing the computational complexity during the actual encoding process, thereby maintaining both accuracy and efficiency.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extends the selection dimension from a single immediately preceding picture to multiple previously-encoded pictures by introducing a similarity computation mechanism. This dimensional expansion allows the system to evaluate encoding information across different time points and select the most appropriate reference picture based on texture similarity, improving complexity index accuracy without significantly increasing system complexity.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Loss of time

If intra slices are used to reduce buffering time and delay, then the encoding speed increases, but the image quality degrades when the intra slice period is short and the area of intra slice region is small

Engineering Contradiction:
Improvebuffering timeVSAvoidimage quality
Core Design Contradiction:
Loss of timeVSManufacturing precision

Solution Approach 1:

The patent dynamically adjusts the quantization step for intra slice regions based on the complexity index derived from similar regions in previously-encoded pictures. By changing the quantization parameter adaptively rather than using a fixed value, the system maintains image quality in small intra slice regions while still achieving reduced buffering time through intra slice encoding.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent copies encoding information (code amount and quantization step) from similar regions in previously-encoded pictures to determine the complexity index and quantization parameters for the current intra slice region. This copying approach allows the system to leverage historical encoding data to maintain image quality without requiring additional buffering or processing time.

Inventive Principle:
Principle #26Copying

3Productivity

If the quantization step is determined based on a fixed initial value, then the encoding process is simple and fast, but the code allocation is not optimized and image quality suffers

Engineering Contradiction:
Improveencoding speedVSAvoidimage quality
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent pre-computes and stores encoding information from multiple previously-encoded pictures, including code amounts and quantization steps for both intra slice and non-intra slice regions. This preliminary preparation enables the system to quickly retrieve relevant encoding information and determine optimized quantization steps without adding significant computational overhead during the encoding process, thus maintaining both speed and quality.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses encoding information from previously-encoded pictures as feedback to determine the quantization step for the current picture. By incorporating historical encoding data into the quantization decision-making process, the system optimizes code allocation and improves image quality while maintaining encoding efficiency through the use of pre-computed reference data.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS9906803B2Quantization control apparatus and method, and quantization control program
Publication Date: 2018.02.27 NIPPON TELEGRAPH & TELEPHONE CORP
  • US9906803B2 patent drawing
  • US9906803B2 patent drawing
  • US9906803B2 patent drawing

AI summary

A quantization control apparatus provided in a video encoding apparatus which utilizes intra slices. The quantization control apparatus includes a similarity computation device that computes similarity between an intra slice region of an encoding target picture and an intra slice region of each of previously-encoded pictures, where the similarity is an index determined such that the closer the images of the two different intra slices, the higher the similarity; a picture selection device that selects a picture having the highest similarity from among the previously-encoded pictures, based on the similarity computed by the similarity computation device; and a quantization step determination device that determines a quantization step used for encoding the encoding target picture, based on encoding information for the picture selected by the picture selection device.