Panoramic Vision System Portrait Screen Distortion Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current panoramic vision systems for vehicles face challenges in displaying a clear and undistorted panoramic view on portrait-format screens due to the inherent distortion of wide-angle cameras, particularly when the image width exceeds the screen width, leading to cropped sides and reduced resolution quality.
Innovation Solution
A panoramic vision system utilizing three cameras with adapted fields of view, an image processing unit that crops and rotates images to fit a portrait-format screen, and correction methods such as projection onto a hemisphere and three-part reconstitution to correct distortion and aggregate images effectively across the screen.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of moving object
If four fisheye cameras are used to capture 360° vision, then the field of view is improved, but image distortion at the periphery worsens
Solution Approach 1:
The panoramic image is divided into three separate zones (left, center, right), each captured by a dedicated camera optimized for its specific field of view. This segmentation allows each camera to operate in its optimal range, reducing peripheral distortion while maintaining comprehensive coverage.
Solution Approach 2:
The central camera acts as an intermediary that captures the central zone where all three cameras overlap. This intermediary view provides a reference for stitching and distortion correction, enabling the left and right cameras to be properly aligned without suffering from extreme peripheral distortion.
2Area of moving object
If the panoramic view width exceeds the portrait screen width, then the field of vision is improved, but the image resolution quality worsens due to cropping
Solution Approach 1:
The panoramic image is segmented into three vertical zones that correspond to the three cameras. Each zone is independently processed and displayed in its appropriate region on the portrait screen, eliminating the need to crop the entire image and preserving resolution quality in each segment.
Solution Approach 2:
The system transitions from a traditional horizontal panoramic display to a vertical portrait orientation by redistributing the three camera zones vertically. This dimensional change allows the full panoramic content to fit on portrait screens without cropping, as the three zones are arranged to utilize the vertical screen real estate efficiently.
3Area of moving object
If three cameras with overlapping fields of view are used, then the image coverage is improved, but the image processing complexity worsens
Solution Approach 1:
The processing system segments the panoramic image into three distinct zones corresponding to each camera's field of view. By processing each zone separately rather than attempting to stitch a single continuous panoramic image, the computational complexity is significantly reduced while maintaining complete coverage.
Solution Approach 2:
Each camera zone is processed with local optimization tailored to its specific characteristics. The central zone receives enhanced processing attention as it contains the overlapping reference information, while the left and right zones are processed according to their respective geometric transformations, reducing overall processing complexity through localized optimization.
Data Source
Figure 1~2b
Figure 3
Figure 4~6
AI summary
The present invention relates to a panoramic vision system (SVP), to a motor vehicle (1) equipped with such a panoramic vision system (SVP) and to an associated displaying method. The panoramic vision system (SVP) comprises three cameras (CC, CLG, CLD) covering a panoramic view, an image-processing unit (UT) and a screen (10) in portrait format able to display the panoramic view captured by the three cameras (CC, CLG, CLD). To do so, the captured images (S10) are reframed (S30) so that one strip (BHG, BHD, BVC) of each image is displayed on the screen (S70), thus reconstructing the panoramic view. Advantageously, the images are corrected for the distortion of the cameras and transposed (S20) using two methods: a so-called "projection onto a cylinder" method and a so-called "reconstruction in three parts" method.