Depth Encoding Quantization Function for 3D Scene Precision

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current technologies face challenges in encoding and decoding depth information of 3D scenes, particularly when the range of depth values is large and the bit depth available for encoding is insufficient, leading to precision loss and visual artefacts during compression and decompression.

Innovation Solution

A method that quantizes depth values using a quantization function defined by a third point, a given angle, and an error value, ensuring that the error angle remains within a predetermined threshold, and encodes this information in a data stream with associated metadata, allowing for accurate de-quantization and reconstruction of 3D scenes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If a fixed quantization step is used for depth encoding, then the encoding process is simple, but precision is lost in regions with large depth variations

Engineering Contradiction:
Improveencoding simplicityVSAvoiddepth precision
Core Design Contradiction:
Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent applies dynamics by making the quantization step adaptive rather than fixed. The quantization step varies dynamically based on the local depth gradient in different regions of the point cloud. In regions with large depth variations, a smaller quantization step is used to maintain precision, while in regions with small depth variations, a larger quantization step is used to reduce data size. This resolves the contradiction by allowing the encoding process to be simple in some regions while maintaining high precision in others.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent applies local quality by using different quantization steps for different regions of the point cloud based on their specific characteristics. The depth gradient is calculated for each region, and the quantization step is adjusted locally according to the gradient magnitude. This ensures that each region is encoded with the appropriate precision for its local depth variations, resolving the contradiction between simplicity and precision by applying quality adaptation locally rather than globally.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If more bits are allocated for depth encoding, then precision is improved, but the data stream size increases

Engineering Contradiction:
Improvedepth precisionVSAvoiddata stream size
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent applies parameter changes by dynamically adjusting the quantization step parameter based on the local depth gradient. Instead of using a fixed number of bits for all depth values, the system changes the quantization parameter (step size) according to the local geometry. This allows the data stream size to be reduced in regions where high precision is not needed while maintaining precision where it is critical, thus resolving the contradiction between precision and data size.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent applies local quality by allocating different precision levels to different regions of the point cloud based on their depth gradient characteristics. Regions with high depth gradients receive finer quantization (more effective precision), while regions with low gradients use coarser quantization. This local adaptation reduces the overall data stream size while maintaining necessary precision in critical areas, resolving the contradiction between precision and data quantity.

Inventive Principle:
Principle #3Local quality

3Measurement precision

If a variable quantization step is used based on depth gradient, then precision is maintained in critical regions, but the encoding complexity increases

Engineering Contradiction:
Improvedepth precision in critical regionsVSAvoidencoding complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by calculating the depth gradient and determining the appropriate quantization step before the actual quantization process. The encoder first computes the gradient magnitude for each region, selects the quantization step based on predefined criteria, and then applies the quantization. This preliminary preparation allows the main encoding process to follow a systematic procedure, managing complexity through structured pre-processing rather than complex real-time decision-making during quantization.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent applies dynamics by implementing a systematic method for adapting the quantization step based on depth gradient. The process dynamically selects from a set of predefined quantization steps according to the local gradient magnitude, creating a controlled dynamic system. This approach manages encoding complexity by using a finite set of adaptive parameters rather than continuous complex calculations, resolving the contradiction between precision and complexity through structured adaptivity.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS12143633B2Method and apparatus for depth encoding and decoding
Publication Date: 2024.11.12 INTERDIGITAL VC HOLDINGS INC
  • US12143633B2 patent drawing
  • US12143633B2 patent drawing
  • US12143633B2 patent drawing

AI summary

Methods, device and data stream format are disclosed in the present document for the encoding, the formatting and the decoding of depth information representative of a 3D scene. Compression and decompression of quantized values by a video codec leads to a value error. This error on values is particularly sensitive for depth encoding. The present invention proposes to encode and decode depth with a quantization function that minimize an angle error when a value error on quantized depth creates a location delta between the projected point and the de-projected point. The inverse of such a quantization function has to be encoded in metadata associated with the 3D scene, for example as a LUT, to be retrieved at the decoding, as such functions are not tractable.