Selective Vehicle Camera Image Extraction with Characteristic Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting image data from a data stream require specific input data for vector comparison, limiting efficiency and flexibility.
Innovation Solution
A computer-implemented method using machine learning algorithms to generate characteristic vectors for each object in a data stream, aggregate these vectors into individual image vectors, and store them in a data memory based on predefined criteria, allowing for efficient and criteria-based extraction of image data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If specific input data are used for comparison of aggregated vectors, then the existing method can determine similarity measures, but it requires specific input data that limits flexibility and efficiency
Solution Approach 1:
The patent extracts only the necessary characteristic vectors from images instead of requiring entire images or specific input data. By generating characteristic vectors through machine learning algorithms and comparing these extracted features, the system achieves flexibility without needing specific input data formats, thus resolving the contradiction between adaptability and complexity
Solution Approach 2:
The patent transforms image data into characteristic vectors by changing the parameter representation from raw pixel data to extracted features. This parameter transformation allows comparison based on meaningful characteristics rather than specific input data requirements, improving flexibility while reducing the complexity of data handling
2Quantity of substance
If all image data from the data stream are stored, then complete data availability is achieved, but data storage requirements and processing time increase significantly
Solution Approach 1:
The patent extracts only relevant images based on comparison criteria between characteristic vectors and previously stored vectors. Instead of storing all images from the data stream, the system selectively extracts and stores only those images that meet predefined criteria, thus maintaining data availability for relevant cases while significantly reducing storage requirements and processing time
Solution Approach 2:
The patent applies partial action by comparing and storing only a subset of images that satisfy specific criteria rather than processing and storing the entire data stream. This selective approach ensures sufficient data availability for training and testing purposes while minimizing the time and resources required for data handling
Data Source
AI summary
A computer-implemented method for the criteria-based extraction of image data, in particular individual images, from a data stream of image data recorded by a camera sensor of a motor vehicle. An aggregation of the characteristic vectors of the particular individual image forms a first individual image vector. A comparison is made of the first individual image vector with plurality of second individual image vectors stored in a data memory. A storage of the first individual image vector and/or the individual image belonging to the first individual image vector is made in the data memory depending on a fulfillment of at least one predefined comparison criterion. A system for the criterion-based extraction of image data is also provided.