Vehicle Surroundings Image Stitching for Moving Object Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques struggle to reliably combine partial images from different perspectives of a vehicle's surroundings, particularly when objects behind the vehicle move relative to the vehicle, leading to incomplete or distorted representations.
Innovation Solution
A method and device that adjust the second partial image based on the object's distance from the vehicle, ensuring seamless stitching and accurate representation of objects across the combined image, using sensors and a processing unit to align and scale images from side and rear cameras.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If images from multiple cameras are merged to provide a comprehensive view of vehicle surroundings, then the coverage area and detection capability are improved, but the processing time and computational complexity increase
Solution Approach 1:
The patent divides the image merging process into distinct processing stages: receiving individual camera images, determining transformation matrices for each image, applying transformations to align images, and finally merging them. This segmentation allows parallel processing of transformation determinations for different cameras, reducing overall processing time while maintaining comprehensive coverage.
Solution Approach 2:
The system pre-determines transformation matrices for each camera image based on known camera positions and orientations before merging. By preparing these transformation parameters in advance, the actual merging operation can proceed more quickly without requiring complex real-time calculations, thus reducing processing time while maintaining accurate alignment.
2Reliability
If images are transformed and merged in real-time during vehicle operation, then the reliability and responsiveness are improved, but the computational load and processing complexity increase
Solution Approach 1:
The system uses pre-stored camera position and orientation data to automatically determine transformation matrices without requiring external intervention or complex real-time calibration. Each camera's intrinsic parameters and extrinsic transformation data are self-contained, allowing the merging process to proceed autonomously with reduced computational complexity while maintaining high reliability through consistent, repeatable transformations.
Solution Approach 2:
The patent transforms images using predetermined transformation matrices that account for camera positions, orientations, and field-of-view parameters. By changing the parameter representation from raw pixel coordinates to transformed coordinate systems based on known camera geometries, the system simplifies the merging process and reduces processing complexity while ensuring accurate spatial alignment and reliable results.
3Measurement precision
If multiple camera images are processed and merged, then the measurement precision and situational awareness are improved, but the data processing time and system complexity increase
Solution Approach 1:
The patent introduces transformation matrices as intermediary elements that mediate between individual camera images and the final merged image. These matrices serve as mathematical intermediaries that encode camera positions, orientations, and geometric relationships, allowing precise alignment without requiring direct complex processing between all image pairs. This intermediary approach maintains high measurement precision while reducing overall system complexity.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
The invention relates to a method (200) for merging partial images of surroundings of a vehicle (105) to form a whole image (305), the method (200) comprising the following steps: capturing (205) a first partial image in a first region of the surroundings, the first region being laterally behind the vehicle (105); capturing (210) a second partial image in a second region of the surroundings, the second region being behind the vehicle (105); detecting (215) an object (110), at least one section of which is in one of the regions; determining (225) a distance between the object (110) and the vehicle (105); adjusting (230) the second partial image depending on the distance; and merging (235) the partial images to form a whole image (305).