Road Image Reconstruction Using Segmentation and Void Filling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing road imagery may suffer from poor quality due to sensor limitations, weather conditions, and occlusions, leading to inaccurate road images that affect navigation and autonomous driving systems.
Innovation Solution
A method involving instance segmentation, void filling, and semantic division to generate accurate road images by identifying and filling void regions, determining common elements, and editing road instances based on semantic information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If road imagery is acquired from sensors, then road images can be obtained, but the quality and accuracy are poor due to sensor limitations, weather conditions, and occlusions
Solution Approach 1:
The patent applies instance segmentation to divide the road image into multiple independent road instances (e.g., different road segments, intersections, or road elements). Each instance can be processed and corrected independently, allowing the system to handle occlusions and weather effects locally without affecting the entire image quality
Solution Approach 2:
The patent performs void filling on road surface images before generating final road images. This preliminary action fills in missing or corrupted regions caused by occlusions or poor sensor data, ensuring complete and accurate road information is available before subsequent processing steps
2Measurement precision
If instance segmentation is performed to identify road instances, then road image accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent uses instance segmentation to divide the road image into multiple independent road instances. This segmentation allows the system to process and correct each instance separately, improving overall accuracy while managing complexity through modular processing
Solution Approach 2:
The patent introduces a void filling step as an intermediary process between road surface image generation and final road image creation. This intermediary step fills missing regions and prepares the data for subsequent processing, simplifying the overall workflow by handling data completeness issues centrally
3Stability of the object's composition
If void filling is performed on road surface images, then integrity and continuity of road instances are ensured, but processing time increases
Solution Approach 1:
The patent performs void filling as a preliminary action on road surface images before generating final road images. By completing this data preparation step early in the pipeline, the system ensures continuous and complete road instance information is available for subsequent processing, avoiding time-consuming corrections later
Solution Approach 2:
The patent creates a void-filled version of the road surface image that serves as a complete reference copy. This copied structure with filled voids can be used repeatedly during road image generation, eliminating the need to reprocess missing regions and reducing overall processing time
Data Source
AI summary
A road image generating method, executed by a computer device, includes obtaining road imagery including a captured or recorded image of a road, a street, or a highway; performing instance segmentation on the road imagery to obtain an instance segmentation result, and generating, based on the instance segmentation result, a road surface image; performing void filling on a void region in the road surface image to obtain a void-filled road surface image, and instantiating the void-filled road surface image to obtain an instantiated image; determining, based on edges of a first plurality of road instances in the instantiated image, a common element shared by the first plurality of road instances; performing, for a road instance, semantic division on the common element to obtain semantic information of the common element; and generating a road image by editing the instantiated image based on the semantic information.