Image Classification Using Range Map Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image classification methods struggle to accurately distinguish between objects with similar attributes due to the lack of utilization of range information, which is not typically available in digital images.
Innovation Solution
A method that uses a digital image processor to receive and classify digital images by integrating range information, representing distances between scene elements and a reference location, to improve image classification accuracy and differentiate between similar visual elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image classification methods are used without range information, then the classification process is simpler and faster, but the accuracy of distinguishing between objects with similar attributes deteriorates
Solution Approach 1:
The patent merges conventional image classification methods with range information processing by integrating multiple data sources (image data, range map data, depth information) into a unified classification system. This combination allows the system to leverage both visual characteristics and spatial information to improve classification accuracy while maintaining a coherent system architecture.
Solution Approach 2:
The patent adds a new dimension to image classification by incorporating range information and depth data alongside traditional 2D image analysis. This dimensional expansion from purely visual features to spatio-visual features enables the system to distinguish between objects with similar attributes by considering their spatial relationships and distances from the camera.
2Measurement precision
If range information is integrated into image classification, then the ability to distinguish between similar visual elements improves, but the computational complexity and processing requirements increase
Solution Approach 1:
The patent segments the classification process into distinct modules: image data processing, range map processing, feature extraction from multiple sources, and integrated classification. This segmentation allows each component to be optimized independently and enables parallel processing of different data types, reducing the computational burden on any single processing unit.
Solution Approach 2:
The patent performs preliminary processing of range information and image data separately before integration, including generating range maps, extracting features from depth information, and pre-processing image data. This preliminary action prepares the data in advance, reducing the computational complexity required during the actual classification phase.
3Reliability
If range information is used in classification, then images with similar visual characteristics can be differentiated, but the data processing requirements and storage needs increase
Solution Approach 1:
The patent extracts only the essential range information and depth features needed for classification rather than processing and storing all raw sensor data. By extracting key spatial characteristics and spatial relationships, the system maintains high classification reliability while reducing the volume of data that needs to be processed and stored.
Solution Approach 2:
Instead of processing complete high-resolution range maps and depth images, the patent inverts the approach by extracting and processing only the essential spatial features and relationships needed for classification. This inversion from processing all data to processing only essential features reduces data volume while maintaining classification reliability.
Data Source
AI summary
A method of identifying an image classification for an input digital image comprising receiving an input digital image for a captured scene; receiving a range map which represents range information associated with the input digital image, wherein the range information represents distances between the captured scene and a known reference location; identifying the image classification using both the range map and the input digital image; and storing the image classification in association with the input digital image in a processor-accessible memory system.