Image Processing Device Litter Shadow Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies for digital cameras face challenges in efficiently and accurately detecting and correcting litter shadows, often requiring substantial processing capabilities and leading to erroneous detections due to changes in brilliance caused by subject coloring, and struggle with accurately identifying foreign objects on optical components.
Innovation Solution
An image processing device comprising a brilliance gradient computation unit, color phase value computation units, and a region extraction unit that computes and compares brilliance and color phase values within and surrounding candidate regions to accurately identify and extract litter shadows and foreign objects, using inverse-gamma-corrected image information and HSV color system for efficient processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If litter shadow correction is performed based on standard image using multiple images, then correction accuracy is improved, but processing complexity and processing time increase substantially
Solution Approach 1:
The invention extracts only the necessary brightness information from a single image to detect litter shadows, rather than processing multiple images. The brightness computation unit calculates brightness values directly from the captured image data, and the litter shadow detection unit identifies shadows by comparing local brightness variations, thereby extracting only the essential information needed for detection without the complexity of multi-image processing
Solution Approach 2:
The invention segments the image processing into distinct functional units: a brightness computation unit that calculates brightness values, and a litter shadow detection unit that detects shadows based on brightness variations. This segmentation allows each unit to perform its specific function efficiently, reducing overall processing complexity while maintaining detection accuracy
2Productivity
If litter shadow is extracted based on brilliance information change, then extraction speed is improved, but erroneous detection increases due to subject coloring changes
Solution Approach 1:
The invention applies local quality analysis by computing brightness values for specific regions and comparing them to determine litter shadows. The litter shadow detection unit analyzes local brightness variations in different regions of the image, comparing brightness values between adjacent regions to identify shadows while being insensitive to global color changes. This local comparison approach maintains extraction speed while improving reliability by focusing on relative brightness differences rather than absolute brilliance values
Solution Approach 2:
The invention changes the parameter used for detection from brilliance information (which is sensitive to color changes) to brightness information (which represents luminance). By computing brightness values that are less sensitive to color variations and more focused on luminance differences, the system maintains extraction speed while reducing erroneous detections caused by subject coloring changes
3Productivity
If conventional brightness-based detection is used, then processing load is reduced, but detection accuracy decreases due to inability to distinguish litter shadows from subject variations
Solution Approach 1:
The invention implements feedback by comparing brightness values between different regions and using this comparison information to refine litter shadow detection. The litter shadow detection unit receives brightness values from the brightness computation unit and uses the differences between adjacent region brightness values to identify litter shadows, providing feedback that improves detection precision while maintaining processing efficiency
Data Source
AI summary
An image processing device comprises: a brilliance gradient computation unit computing a brilliance gradient within an image based on an image information; a color phase value computation unit computing a first color phase value and a second color phase value, the first color phase value representing a color phase inside a candidate region determined based on the brilliance gradient computed by the brilliance gradient computation unit, the second color phase value representing a color phase in a region surrounding the candidate region; and a region extraction unit extracting a region from the candidate region such that a difference between the first color phase value and the second color phase value is less than or equal to a predetermined threshold value.