Overhead View Image Boundary Discontinuity Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems for vehicles face challenges in creating seamless overhead-view images due to production errors in imaging devices, leading to discontinuities and potential confusion for drivers, especially when objects are displayed double around boundaries, affecting safety.
Innovation Solution
An image processing apparatus that detects object positions and assigns evaluation values to images from different directions, determining boundaries to ensure objects are included in the image with the higher evaluation value, thereby minimizing discontinuities and maintaining object continuity within a single image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If images from multiple imaging devices are joined along boundaries to create an overhead-view image, then all objects in the region around the vehicle can be captured, but discontinuities occur at the boundaries due to production errors in the imaging devices
Solution Approach 1:
The patent applies parameter changes by adjusting the transparency parameter of images in overlapping regions. Specifically, when an object is detected in an overlapping region, the transparency of the image containing that object is reduced (made more translucent) to allow seamless blending with adjacent images, thereby eliminating boundary discontinuities while maintaining object visibility across the entire overhead-view image
2Manufacturing precision
If images are made translucent and overlapped around boundaries to reduce discontinuity, then visibility is improved, but objects may be displayed double causing confusion for the driver
Solution Approach 1:
The patent employs feedback by using object detection results to control image transparency dynamically. The detecting unit identifies objects in overlapping regions, and this detection information feeds back to the generating unit, which then adjusts the transparency of specific images based on the detected object positions. This closed-loop control ensures that objects are displayed clearly without duplication, as the transparency adjustment is precisely targeted to regions containing detected objects
Solution Approach 2:
The patent applies local quality by making different parts of the composite image have different transparency properties. Specifically, only the regions containing detected objects in overlapping areas are made translucent, while other regions maintain their original opacity. This localized approach ensures that objects are clearly visible without being duplicated, while still achieving seamless blending at boundaries where needed
3Area of stationary object
If multiple imaging devices are equipped at front, rear, left, and right sections to image all objects, then complete coverage is achieved, but the complexity of the system increases
Solution Approach 1:
The patent applies segmentation by dividing the overall image processing task into distinct functional units: a detecting unit for identifying objects in overlapping regions, and a generating unit for composing the final overhead-view image with appropriate transparency adjustments. This functional segmentation allows the system to handle complex multi-device image processing through modular, manageable components, reducing the perceived complexity while maintaining complete coverage
Data Source
AI summary
An image processing apparatus: detects an object position included within an overlapping region of a first image corresponding to a vehicles's traveling direction and a second image corresponding to a direction crossing the traveling direction; assigns the first image a first value when the object is in a first position, and a second value smaller than the first value when the object is in a second position more distant from a traveling direction axis of the vehicle than the first position, and assigns the second image a third value when the object is in the first position, and a fourth value larger than the third value when the object is in the second position; and determines a boundary so that the object is included within an image having a larger value than the other; and joins the first and second images.