Stereo Camera Parallax Calculation Using Background Degree Weighting

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Stereo camera devices face challenges in accurately distinguishing objects from their background at greater distances, leading to increased parallax errors and reduced distance measurement accuracy.

Innovation Solution

The method involves extracting one image object region from a pair of images, calculating the likelihood of background configuration parts, and then extracting a similar object region from the other image, applying weights based on the background degree to enhance parallax calculation accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Length of stationary object

If stereo camera device measures distance to distant objects, then measurement range is extended, but distance accuracy deteriorates due to increased parallax error and coarser image quality

Engineering Contradiction:
Improvemeasurement rangeVSAvoiddistance accuracy
Core Design Contradiction:
Length of stationary objectVSMeasurement precision

Solution Approach 1:

The patent extracts and processes only the object region of interest from the full image, separating it from the background. By calculating the degree of background for each pixel and applying weights to exclude background areas, the system focuses computational resources on the relevant object regions, thereby improving parallax calculation accuracy for distant objects without requiring full-image processing

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies different processing qualities to different regions of the image. Specifically, it calculates background degree and applies weighted processing selectively to object regions versus background regions. This local differentiation allows the system to maintain high measurement precision for distant objects by concentrating processing power where it is most needed, rather than uniformly processing the entire image

Inventive Principle:
Principle #3Local quality

2Measurement precision

If full image processing is performed to improve object-background distinction, then measurement accuracy improves, but computational load and processing time increase

Engineering Contradiction:
Improveobject-background distinction accuracyVSAvoidprocessing efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent extracts only the necessary object regions from the full image by calculating background degree for each pixel. This extraction process identifies and isolates regions containing objects of interest, excluding background areas from detailed processing. Consequently, the system achieves accurate object-background distinction while significantly reducing the computational load compared to processing the entire image

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies partial processing to the image by performing background degree calculation and weighted processing only on relevant object regions rather than the entire image. This partial action approach maintains measurement precision for object detection while avoiding the excessive computational burden of processing all image pixels uniformly, thus improving processing efficiency

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP2778603B1Image processing apparatus and image processing method
Publication Date: 2017.06.21 HITACHI AUTOMOTIVE SYST LTD
  • EP2778603B1 patent drawingFigure 1
  • EP2778603B1 patent drawingFigure 2~3
  • EP2778603B1 patent drawingFigure 4~4(b)

AI summary

To obtain an image processing device and a method of processing an image that improve distance accuracy and is capable of performing accurate distance even about an object at a greater distance than before, when a distance to an object is measured, one image object region 302 including an image of an object is extracted from one image of a pair of images imaged by a pair of imaging elements at the same time in the same direction. The degree of background that is likelihood of whether either an object image configuration part 304 or a background image configuration part 303 is calculated for each of a plurality of image configuration parts that configures the one image object region 302. Then, the other image object region 503 having an image similar to the one image object region 302 is extracted from the other image 501 using the degree of background, and a parallax between the one image object region 302 and the other image object region 503 is calculated.