Image Recognition Using Auto-Exposure Data for Subregion Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image recognition technologies using semantic segmentation lack effective methods to accurately recognize regions in images using subsidiary information obtained during image capture, such as auto-exposure data, which is crucial for precise image segmentation and object recognition.
Innovation Solution
An image recognition apparatus comprising an acquiring unit for obtaining image data and subsidiary information, a segmenting unit for segmenting the image into subregions based on the acquired subsidiary information, and a determination unit for categorizing these subregions using extracted feature values, specifically employing evaluation values for automatic focusing, auto-exposure, and automatic white balance to enhance segmentation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If semantic segmentation is performed using only image data without subsidiary information, then the processing is simpler, but the recognition accuracy of regions is insufficient
Solution Approach 1:
The patent applies preliminary action by acquiring subsidiary information (auto-exposure evaluation values) before performing semantic segmentation. This preliminary acquisition of capture condition data enables more accurate region recognition during the segmentation process, as the auto-exposure values provide prior knowledge about lighting conditions that help distinguish regions more effectively.
Solution Approach 2:
The patent uses auto-exposure evaluation values as an intermediary element between the image data and the segmentation process. These evaluation values serve as mediator information that bridges the gap between raw image data and accurate region recognition, providing additional contextual information about capture conditions without requiring complex multi-sensor systems.
2Measurement precision
If conventional file format images (JPEG, BMP) are used, then storage space is efficient, but sufficient information for highly accurate image recognition is not provided
Solution Approach 1:
The patent applies segmentation by dividing the image data into multiple regions based on auto-exposure evaluation values. This segmentation approach extracts meaningful spatial information from the image while utilizing subsidiary information to guide the segmentation process, thereby achieving accurate region recognition without requiring storage of the entire high-resolution image data.
Solution Approach 2:
The patent extracts essential feature information from both the image data and the subsidiary auto-exposure information. By taking out only the critical features needed for region recognition rather than processing the complete image data, the system achieves high recognition accuracy while maintaining efficient information usage.
3Reliability
If subsidiary information from image capture is integrated into semantic segmentation, then region categorization accuracy is improved, but the processing complexity increases
Solution Approach 1:
The patent applies parameter changes by utilizing auto-exposure evaluation values (numerical parameters representing capture conditions) to adjust and refine the semantic segmentation process. These parameter changes enable the system to adapt the segmentation thresholds and region boundaries based on the specific lighting and capture conditions, thereby improving categorization accuracy without requiring a complete redesign of the segmentation architecture.
Data Source
AI summary
On the basis of subsidiary information associated with image data, an image for the image data is segmented into multiple subregions, and feature values are extracted for each of the subregions obtained through the segmentation. The category for each of the subregions is determined on the basis of the extracted feature values.