Hyper-spectral Image Analysis via Dynamic Database Updating
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current hyper-spectral imaging and analysis methods face challenges in achieving high accuracy, precision, and speed simultaneously, particularly in distinguishing objects of interest from a large number of objects of non-interest, especially when the ratio of objects of interest is very low, and in processing large volumes of data at high speeds.
Innovation Solution
A method involving dynamic database updating, where a first reference object database of hyper-spectral image data and a second reference database of biological and physical data are used to identify and classify objects, allowing for real-time or offline processing and analysis, integrating data manipulation and analysis operations with high accuracy, precision, and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hyper-spectral imaging and analysis is performed with high accuracy and precision, then the ability to distinguish objects of interest from non-interest is improved, but the processing time and speed are reduced
Solution Approach 1:
The patent segments the hyper-spectral data processing into multiple stages: initial rapid classification to identify potential objects of interest, followed by detailed analysis only for those identified objects. This segmentation allows the system to process large datasets quickly while maintaining high accuracy for objects of interest by applying comprehensive analysis only where needed.
Solution Approach 2:
The patent implements partial action by performing full-spectrum detailed analysis only on a subset of objects identified as potential objects of interest, rather than analyzing every object in the dataset with the same level of detail. This approach maintains high distinguishing accuracy for target objects while significantly reducing overall processing time.
2Measurement precision
If the ratio of objects of interest to total objects is very low, then the challenge of identifying objects of interest increases, but the processing time for the entire dataset increases significantly
Solution Approach 1:
The patent applies preliminary action through a two-stage process: first performing a rapid initial classification to identify potential objects of interest from the entire dataset, then performing detailed analysis only on those identified candidates. This preliminary filtering action dramatically reduces processing time for low-ratio datasets while maintaining high detection capability for the rare objects of interest.
Solution Approach 2:
The patent introduces an intermediary classification stage that acts as a mediator between the full dataset and the detailed analysis phase. This intermediary step quickly identifies potential objects of interest, allowing the system to handle low ratios efficiently by focusing computational resources only on relevant candidates rather than processing every object in detail.
3Productivity
If large volumes of hyper-spectral data are processed at high speeds, then productivity is improved, but the accuracy and precision of analysis may be reduced
Solution Approach 1:
The patent segments the analysis process into rapid initial classification and detailed subsequent analysis, allowing high-speed processing of large data volumes through the first stage while maintaining high accuracy through the second stage for identified objects of interest.
Solution Approach 2:
The patent applies local quality by providing different levels of analysis quality to different data elements: rapid classification for all objects and detailed analysis only for objects of interest. This ensures high productivity overall while maintaining high analysis accuracy where it matters most.
Data Source
AI summary
Processing and analyzing hyper-spectral image data and information via dynamic database updating. (a) processing/analyzing representations of objects within a sub set of the hyper spectral image data and information, using a first reference database of hyper spectral image data, information, and parameters, and, a second reference database of biological, chemical, or/and physical data, information, and parameters. Identifying objects of non-interest, and objects of potential interest, from the data/information sub-set. (b) processing/analyzing identified objects of potential interest, by further using first and second reference databases. Determining absence or presence of objects of interest, additional objects of non-interest, and non-classifiable objects of potential interest, from the data/information sub set. (c) updating first and second reference databases, using results of (a) and (b), for forming updated first and second reference databases. (d) repeating (a) through (c) for next sub-set of hyper spectral image data/information, using updated first and second reference databases. (e) repeating (d) for next sub-sets of hyper spectral image data/information.