Road Dividing Object Detection via Region Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional road dividing object detection methods are complex and time-consuming, focusing on pixel-level processing which hampers the accuracy and efficiency required for driving assistance systems.

Innovation Solution

The method detects road dividing object regions instead of individual pixels, using feature points to create a middle-layer expression, reducing the complexity of modeling and post-processing algorithms, and improving detection accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If pixel-level processing is used for road dividing object detection, then detection detail is improved, but system complexity and processing time increase

Engineering Contradiction:
Improvedetection detailVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the image processing task into two levels: region-level detection using Hough transform to identify road dividing object regions, and point-level detection within those regions to locate specific feature points. This hierarchical segmentation reduces overall system complexity while maintaining detection precision.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from pure pixel-level processing to a multi-dimensional approach by first detecting road dividing object regions (spatial dimension) and then detecting feature points within those regions. This dimensional transition reduces computational complexity while preserving detection accuracy.

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

2Measurement precision

If pixel-level processing is used for road dividing object detection, then detection detail is improved, but processing speed decreases

Engineering Contradiction:
Improvedetection detailVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent divides the detection process into two stages: first detecting road dividing object regions using Hough transform, then detecting feature points only within those identified regions. This segmentation reduces the total number of pixels requiring detailed processing, thereby improving processing speed while maintaining detection precision.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a regional dimension to the detection process, moving from direct pixel-level analysis to region-based analysis followed by point-level analysis. This dimensional approach reduces computational load and improves processing speed without sacrificing detection detail.

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

3Measurement precision

If complicated road dividing object modeling and post-processing are used, then detection accuracy is improved, but algorithm complexity increases

Engineering Contradiction:
Improvedetection accuracyVSAvoidalgorithm complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the detection task into region detection (using Hough transform) and feature point detection (within regions). This segmentation eliminates the need for complicated overall modeling and post-processing while maintaining high detection accuracy through the focused point-level analysis within identified regions.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3079099B1Method and device for detecting road dividing object
Publication Date: 2020.03.11 RICOH CO LTD
  • EP3079099B1 patent drawingFigure 1
  • EP3079099B1 patent drawingFigure 2
  • EP3079099B1 patent drawingFigure 3

AI summary

Disclosed is a method of detecting at least one road dividing object. The method comprises a first obtainment step of obtaining an image including a road region; a region detection step of detecting at least one road dividing object region in the road region of the image; a feature point detection step of detecting plural feature points of the at least one road dividing object in the at least one road dividing object region; and a determination step of determining, based on the plural feature points of the at least one road dividing object in the at least one road dividing object region, the at least one road dividing object.