Partial Image Generation for Bird's Eye View Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in accurately combining partial images from multiple vehicles into bird's eye images due to variations in camera mounting positions and optical axes, leading to difficulties in generating a unified road image for vehicle self-driving systems.
Innovation Solution
A partial image generating device that assesses whether the road in an image is straight, using the vanishing point or vehicle type to determine the appropriate region for image cutting, and adjusts dimensions based on road surface roughness and vehicle speed to ensure consistent representation of road features across images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the server uses multiple partial images from different vehicles to generate bird's eye images, then the completeness of road information is improved, but the difficulty of combining images due to varying camera positions and optical axes increases
Solution Approach 1:
The patent introduces an assessment unit as an intermediary that evaluates road straightness before image processing. This intermediary component analyzes whether the road in each partial image is straight and determines the appropriate processing method accordingly, mediating between the varied input images and the final combined bird's eye image to resolve the complexity of image combination
Solution Approach 2:
The patent changes processing parameters based on road characteristics. When the road is straight, the system uses vanishing point-based processing; when the road is curved, it uses vehicle-type-based processing. This parameter change approach allows the system to adapt to different road conditions and simplifies the image combination process by selecting appropriate processing methods for each image
2Measurement precision
If the partial image region is determined based on vanishing point for straight roads, then the precision of road representation is improved, but the reliability decreases when roads are curved
Solution Approach 1:
The patent implements a dynamic processing approach where the system first assesses road straightness and then dynamically selects the appropriate processing method. For straight roads, it uses vanishing point-based processing for high precision; for curved roads, it switches to vehicle-type-based processing for reliability. This dynamic adaptation resolves the contradiction between precision and reliability
Solution Approach 2:
The assessment unit performs preliminary action by evaluating road straightness before the actual image processing. This preliminary assessment determines which processing method to apply, ensuring that the most appropriate method is selected in advance, thereby maintaining both precision when applicable and reliability across all conditions
3Area of stationary object
If the second region dimension in widthwise direction is larger, then the coverage of curved roads is improved, but the amount of data to be processed increases
Solution Approach 1:
The patent applies local quality by determining different region dimensions based on local road characteristics. For straight roads, it uses the smaller first region optimized for precision; for curved roads, it uses the larger second region optimized for coverage. This local adaptation ensures that each image region is sized appropriately for its specific context, balancing coverage and data volume
Data Source
AI summary
A partial image generating device has a processor configured to generate a partial image from an image in which the environment surrounding a vehicle has been photographed using an imaging device provided in the vehicle, and to assess whether or not the road represented in the image is straight. The processor is configured so that, when it has been assessed that the road represented in the image is straight, it generates a partial image by cutting out a first region in the image that is estimated to contain the road, based on the vanishing point of the road, and when it has assessed that the road represented in the image is not straight, it generates a partial image by cutting out a second region in the image determined based on the type of the vehicle.