Shadow Model Background Detection for Vehicle Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional background subtraction methods in image processing, particularly in vehicles, often misidentify moving shadows as foreground objects, leading to false detections of foreign matter or people in changing environments, such as when interior changes or stickers are added.
Innovation Solution
An information processing device that captures actual and calibration images, generates a shadow model by extracting shadow regions and performing color tone correction, and creates an estimated background image by adding shadows to the calibration image, simulating the background at the time of the actual image capture while excluding foreign matter.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional background subtraction methods are used to detect background in images, then the detection process is simple, but moving shadows are misidentified as foreground objects leading to false detections
Solution Approach 1:
The system performs preliminary actions by capturing calibration images in advance when no foreign objects are present, and pre-processing actual images to extract shadow regions before background subtraction. This preliminary processing of shadows and calibration data enables accurate background detection without misidentifying moving shadows as foreground objects.
Solution Approach 2:
The system introduces intermediary elements including a shadow extraction model that identifies and processes shadow regions, and a calibration image that serves as a reference for background characteristics. These intermediaries mediate between the raw image data and final background detection, preventing false identification of shadows as foreign objects.
2Productivity
If simple background subtraction is used, then processing is fast, but false alarms occur when interior changes or stickers are added
Solution Approach 1:
The system captures calibration images in advance when the interior is in a known state without foreign objects. This preliminary calibration data is stored and used as a reference for subsequent detections, enabling the system to adapt to interior changes and distinguish between normal interior modifications and actual foreign objects, thereby maintaining detection reliability.
Solution Approach 2:
The system changes detection parameters dynamically by comparing actual images against calibration images and adjusting the background model based on detected shadow regions and interior changes. This parameter adaptation allows the system to maintain high detection reliability even when interior conditions change, such as when stickers are added or lighting conditions vary.
Data Source
AI summary
A control unit executes: acquiring an actual image that is an image captured by the imaging unit; acquiring a calibration image that is an image captured by the imaging unit in the past and serving as a reference for the current background; generating a model for estimating a region where a shadow exists and correcting the color tone; and adding a shadow corresponding to the actual image to the calibration image based on the calibration image and the model and correcting the color tone and generating an estimated background image, which is a corrected image and is a background image corresponding to the actual image.