Intra Prediction for Depth Images Using Boundary Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video processing technologies face inefficiencies in encoding and decoding 3D videos, particularly in predicting pixel values for depth images, which affect the encoding/decoding efficiency of 3D videos.

Innovation Solution

An intra prediction method is developed that determines pixel values in a prediction target block of a depth image by using boundary pixels and chain codes to split the block into regions, allowing for efficient prediction based on neighboring pixels, enhancing encoding/decoding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional intra prediction methods are used for depth images, then encoding/decoding process is simpler, but encoding/decoding efficiency is lower

Engineering Contradiction:
Improveencoding/decoding efficiencyVSAvoidprediction process complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The prediction target block is divided into multiple prediction target regions based on boundary pixels and chain codes. Each region is predicted separately using neighboring pixels specific to that region, allowing the algorithm to adapt to local depth variations while maintaining overall efficiency

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different prediction methods are applied to different regions within the prediction target block. By identifying boundary pixels and creating chain codes, the method applies localized prediction strategies that match the local depth characteristics, improving overall prediction accuracy without uniformly increasing complexity

Inventive Principle:
Principle #3Local quality

2Measurement precision

If simple prediction methods are used, then processing speed is faster, but prediction accuracy is lower

Engineering Contradiction:
Improvepixel value prediction accuracyVSAvoidprediction processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

Boundary pixels are identified and chain codes are generated before the actual prediction process. This preliminary segmentation allows the main prediction algorithm to work more efficiently by operating on pre-defined regions rather than searching for boundaries during prediction

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The prediction method dynamically adapts to the local characteristics of depth images by using chain codes to represent boundary shapes. The prediction process adjusts to follow the actual geometry of depth boundaries, improving accuracy for complex shapes while maintaining speed through algorithmic optimization

Inventive Principle:
Principle #15Dynamics

3Loss of information

If conventional prediction methods are used, then processing is simpler, but compression performance is worse

Engineering Contradiction:
Improvecompression performanceVSAvoidboundary handling complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The method changes the representation parameters by using chain codes to encode boundary pixel sequences. This parameter transformation allows efficient representation of complex boundary shapes, improving compression performance by better capturing the geometric characteristics of depth images

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The boundary information is copied and represented through chain codes that can be efficiently transmitted and reconstructed. This copying mechanism allows the boundary geometry to be preserved with minimal bit rate overhead, improving compression efficiency

Inventive Principle:
Principle #26Copying

Data Source

PatentEP2750384B1Intra prediction method and an apparatus for decoding an image
Publication Date: 2020.07.22 LG ELECTRONICS INC
  • EP2750384B1 patent drawingFigure 1
  • EP2750384B1 patent drawingFigure 2
  • EP2750384B1 patent drawingFigure 3

AI summary

The present invention relates to an intra prediction method for predicting pixel values within a prediction target block on the basis of a plurality of peripheral pixels adjacent to the prediction target block. The intra prediction method according to the present invention includes the steps of: receiving and decoding encoded image information; determining a target boundary by determining a plurality of boundary pixels representing the target boundary within the prediction target block on the basis of the decoded image information; determining a plurality of prediction target regions partitioned on the basis of the target boundary within the prediction target block; and performing a prediction for each of the plurality of prediction target regions on the basis of peripheral pixels differing from each other selected from among the plurality of the peripheral pixels.