Surround-View Image Blending for Clear Overlapping Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional vehicle surrounding area image generation devices blur objects in overlapping imaging areas due to uniform alpha blending, regardless of object presence or distance, resulting in unclear overhead images.
Innovation Solution
A vehicle surrounding area image generation device that determines object distance using sensors and separates overlapping image portions into road surface and object images, allowing for selective alpha blending to maintain object clarity in the generated overhead image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If uniform alpha blending is applied to combine images in overlapping portions, then the image combination process is simple, but objects in the overlapping portions become blurred and the overhead image becomes unclear
Solution Approach 1:
The patent segments the overlapping portion into multiple regions based on object detection results. The processing area is divided into a first region containing detected objects and a second region without objects. Different alpha blending ratios are applied to each region, allowing objects to be preserved clearly while still achieving smooth transitions in background areas.
Solution Approach 2:
The patent applies different alpha blending ratios to different regions within the overlapping portion. The first region (containing objects) uses a higher alpha ratio to preserve object clarity, while the second region (background) uses a lower alpha ratio for smoother blending. This local differentiation resolves the contradiction between simplicity and clarity.
2Productivity
If the alpha value is set regardless of object presence in the overlapping portion, then the image processing is efficient, but objects become blurred in the overhead image
Solution Approach 1:
The patent performs object detection and determines the presence of objects in the overlapping portion before applying alpha blending. This preliminary action allows the system to select appropriate alpha ratios based on object presence, ensuring objects remain clear while maintaining efficient processing through automated detection and conditional processing.
Solution Approach 2:
The patent dynamically adjusts the alpha blending ratio based on object detection results. When objects are detected in the overlapping portion, a higher alpha ratio is applied to preserve object clarity. When no objects are present, a lower alpha ratio is used for smoother blending. This dynamic adjustment maintains both efficiency and clarity.
3Adaptability or versatility
If alpha blending weights images based on distance from the center of the imaging area, then the combination follows a systematic rule, but objects may still become blurred
Solution Approach 1:
The patent overrides the distance-based alpha weighting by applying object-aware regional differentiation. Instead of uniformly applying distance-based weights, the system identifies regions containing objects and applies higher alpha ratios specifically to those regions, ensuring objects remain clear regardless of their position relative to the imaging center.
Solution Approach 2:
The patent uses object detection feedback to adjust the alpha blending strategy. The detection results provide information about object locations, which then feeds into the alpha ratio selection process. This feedback mechanism ensures that objects are preserved clearly while maintaining systematic and adaptive image combination.
Data Source
AI summary
An ECU generates an overhead image including a vehicle and a vehicle surrounding area, on the basis of a plurality of images taken by a plurality of cameras mounted on the vehicle. When a distance between an object existing in an overlapping portion of imaging areas of the plurality of cameras and the vehicle is within a threshold and the object exists in a traveling direction of the vehicle, the overlapping portion is divided into two pairs of a road surface image and an object image, and combination is performed by alpha blending with an α value for the object image being set to 1 and an α value for the road surface image being set to 0 for an area that is the object image in one of the two pairs and is the road surface image in the other of the two pairs, to generate the overhead image.