Overhead Image Generation Plane Segmentation Wall Deformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques fail to accurately transform images captured by in-vehicle devices, as they do not distinguish between road surfaces and walls, leading to deformation of wall images when generating overhead views.
Innovation Solution
An image processing system that acquires information to identify plane positions in captured images, determines correspondence relationships between image regions, and corrects these regions to generate overhead images accurately, distinguishing between different objects like road surfaces and walls.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the same image processing is applied to both road surfaces and walls, then the processing simplicity is maintained, but the shape accuracy of wall images deteriorates
Solution Approach 1:
The captured image is segmented into multiple regions based on plane information, distinguishing between road surfaces and walls. Each region is then processed independently with appropriate transformation parameters, allowing road surfaces to be transformed with one set of parameters and walls with another, thereby maintaining shape accuracy for both types of objects
Solution Approach 2:
Different transformation parameters and processing methods are applied to different regions of the image based on their semantic meaning. Road surface regions receive transformations optimized for ground planes, while wall regions receive transformations optimized for vertical surfaces, ensuring each region maintains its shape accuracy according to its local characteristics
2Manufacturing precision
If plane identification is implemented to distinguish road surfaces and walls, then the shape accuracy of transformed images is improved, but the device complexity increases
Solution Approach 1:
Plane information is acquired and processed in advance before the main image transformation occurs. The system pre-identifies road surface and wall regions, determines their respective plane equations, and prepares transformation parameters beforehand, which streamlines the subsequent image transformation process and reduces real-time computational complexity
Solution Approach 2:
A plane information data structure serves as an intermediary between the raw captured image and the transformation process. This intermediary contains pre-computed plane equations and region masks that guide the transformation, decoupling the complex plane detection task from the image transformation task and making the overall system more manageable
Data Source
AI summary
An image processing apparatus includes an acquisition unit configured to acquire information for identifying a position of a plane in a captured image, where the captured image includes a structure composed of a plurality of planes as an object, a determination unit configured to determine a correspondence relationship between a region corresponding to the plane in the captured image and a region in an overhead image to be created, according to the information acquired by the acquisition unit, and a generation unit configured to correct an image of the region corresponding to the plane in the captured image so that the corrected image corresponds to the region in the overhead image to be created, according to the correspondence relationship determined by the determination unit, and generate an overhead image composed of a plurality of corrected images that are combined with each other.