3D Image Encoder Line-by-Line Lossless Compression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current lossless compression techniques for digital images, particularly 3D images with range data, face challenges in achieving high compression ratios and real-time processing over bandwidth-limited communication links like Gigabit Ethernet, often requiring significant hardware resources and incurring latency.

Innovation Solution

A modified prediction-based encoding method that encodes pixel values independently per line, using a context of two preceding pixels for prediction and adaptively switching to run length mode for missing data, employing Simplified Adjusted Binary Coding (SABC) or Golomb-Rice Coding (GRC) with Run Length Coding (RLC), to achieve efficient lossless compression.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional lossless compression techniques are used for 3D images with range data, then compression is achieved, but processing speed and throughput are limited by hardware resources and latency

Engineering Contradiction:
Improvecompression throughputVSAvoidhardware resources
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent divides the image processing into independent line-by-line operations, where each line is compressed separately using prediction based on previous lines. This segmentation allows parallel processing of multiple lines simultaneously, increasing throughput without requiring proportionally more hardware resources for sequential processing

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent employs dynamic prediction models that adapt to the specific characteristics of range data from 3D imaging. The prediction algorithm dynamically adjusts based on the correlation between adjacent lines, optimizing compression efficiency for real-time processing while maintaining manageable hardware requirements

Inventive Principle:
Principle #15Dynamics

2Quantity of substance

If high compression ratios are achieved through sophisticated algorithms, then data reduction is improved, but processing latency increases

Engineering Contradiction:
Improvedata compression ratioVSAvoidprocessing latency
Core Design Contradiction:
Quantity of substanceVSLoss of time

Solution Approach 1:

The patent performs prediction using data from previous lines before encoding the current line. This preliminary prediction step reduces the entropy of the data early in the process, allowing for more efficient subsequent encoding and reducing overall processing time while achieving high compression ratios

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent maintains continuous processing by using already-compressed data from previous lines as input for predicting the current line. This continuous flow of useful action eliminates idle processing time and reduces latency while sustaining high compression efficiency throughout the entire image processing pipeline

Inventive Principle:
Principle #20Continuity of useful action

3Productivity

If bandwidth-limited communication links are used for image transmission, then network cost is reduced, but image transmission speed is limited

Engineering Contradiction:
Improveimage transmission speedVSAvoiddata volume
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The patent transforms the image data from its original high-bit-depth format into a compressed representation with fewer bits per pixel through prediction-based encoding. This parameter change in data representation reduces the total data volume by a factor of 2-4x while preserving all essential information, enabling faster transmission over bandwidth-limited links

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3734973B1Method and encoder relating to encoding of pixel values to accomplish lossless compression of a digital image
Publication Date: 2023.07.05 SICK IVP
  • EP3734973B1 patent drawingFigure 1A~1B
  • EP3734973B1 patent drawingFigure 2~3
  • EP3734973B1 patent drawingFigure 4

AI summary

Encoder (131; 800) and method for encoding of pixel values of a digital image comprising multiple lines of pixels to accomplish lossless compression of the digital image. For each of said multiple lines the encoder (131; 800) obtains (701) unencoded pixels values of the line. Further, for each of said multiple lines, the encoder (131; 800) determines (702), for each of one or more pixels of the line, which encoding to be used for encoding of the unencoded pixel value of the pixel (x) in said lossless compression of the digital image. The determination being based on how said unencoded pixel value relates to unencoded pixel values of other, closest neighboring pixels (N1, N2) of said line.