Map Data Update via Vehicle Image Registration and Transparency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for creating and updating detailed maps for autonomous vehicles face challenges in accurately reflecting changes in road environments and fixed geographic features, especially when these features are obscured or dynamic.
Innovation Solution
An image processing method and apparatus that utilize images from multiple vehicles to create and update map data by performing image registration, transparency processing, and comparison with stored map data to identify and reflect changes in static objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If images from multiple vehicles are processed to update map data, then the accuracy and timeliness of map data is improved, but the processing complexity and computational load increases
Solution Approach 1:
The patent segments the image processing task into distinct stages: image acquisition from multiple vehicles, image registration to align perspectives, transparency processing to handle overlapping features, static object extraction to identify permanent road features, and comparison with existing map data. This segmentation allows each processing stage to be optimized independently, managing overall system complexity while achieving high-accuracy map updates.
Solution Approach 2:
The patent introduces intermediate processing steps including image registration and transparency processing as mediators between raw vehicle images and final map data. These intermediary processes transform multiple vehicle perspectives into a unified coordinate system and handle occlusions, enabling accurate map updates without requiring direct complex comparison of all raw images simultaneously.
2Measurement precision
If transparency processing is applied to registered images, then static objects can be accurately extracted, but processing time and computational resources increase
Solution Approach 1:
The patent applies transparency processing as a preliminary action before static object extraction. By pre-processing the registered images with transparency operations, the system prepares the data in advance to facilitate easier and more accurate identification of static objects in subsequent steps, reducing the computational burden during the extraction phase itself.
Solution Approach 2:
The patent replaces traditional mechanical or direct comparison methods for identifying static objects with transparency-based image processing. Instead of manually or mechanically analyzing overlapping images, the system uses computational transparency operations to automatically differentiate static road features from dynamic elements, significantly reducing processing time while maintaining accuracy.
3Reliability
If map data is updated frequently with real-time vehicle images, then the map reflects current road conditions accurately, but the system requires continuous data collection and processing
Solution Approach 1:
The patent implements periodic action by collecting images from vehicles at regular intervals and updating map data based on accumulated data rather than continuous real-time processing. This approach maintains map reliability by regularly incorporating new information while reducing energy consumption by processing data in periodic batches rather than continuously, allowing the system to balance accuracy requirements with resource constraints.
Data Source
AI summary
Disclosed is an image processing method. The method includes the steps of receiving an image obtained from a plurality of vehicles positioned on a road, storing the received images according to acquisition information of the received images; determining a reference image and a target image based on images having the same acquisition information among the stored images, performing an image registration using a plurality of feature points extracted from each of the determined reference image and target image, performing a transparency process for each of the reference image and the target image performed with the image registration, extracting static objects from the transparency-processed image, and comparing the extracted static objects with objects on an electronic map pre-stored to updating the electronic map data, when the objects on the electronic map data pre-stored are different from the extracted static objects.