Image Pattern Recognition Using Local Minimum Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pattern recognition methods in image data sets are too time-consuming for real-time applications, such as military reconnaissance, due to the inefficiency in processing large volumes of image data.
Innovation Solution
A method that forms data vectors for each pixel in an image data set using characteristic coefficient values from a test environment, compares these with reference data vectors, and determines pattern presence by calculating difference values and checking for a threshold and local minimum, allowing for faster recognition through rotational invariance and selective data processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional pattern recognition methods are used to process image data sets, then comprehensive pattern detection can be achieved, but the recognition time becomes too long for real-time applications
Solution Approach 1:
The patent divides the image data set into multiple subareas and processes them separately using parallel computation. The image is segmented into regions that can be evaluated independently, allowing simultaneous processing of multiple areas without requiring sequential analysis of the entire image, thus reducing total recognition time while maintaining detection accuracy.
Solution Approach 2:
The patent performs preliminary classification of subareas based on simple criteria before applying full pattern recognition algorithms. By pre-identifying regions of interest using basic features, the system avoids computationally intensive processing of entire images, reducing recognition time while ensuring that potential patterns are not missed.
2Reliability
If the entire image data set is processed for pattern recognition, then complete coverage is achieved, but processing speed decreases
Solution Approach 1:
The image data set is divided into multiple subareas that are processed in parallel. This segmentation allows the system to maintain complete coverage of the entire image while processing different regions simultaneously, thereby improving throughput without sacrificing detection completeness.
Solution Approach 2:
The patent applies full pattern recognition algorithms only to selected subareas that meet certain criteria, rather than processing every pixel in the entire image. This partial action approach maintains reliability for regions requiring detailed analysis while improving overall productivity by reducing the scope of intensive processing.
3Measurement precision
If detailed analysis of all image data is performed, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent applies different processing strategies to different subareas based on their characteristics. Regions with high potential for containing patterns receive detailed analysis, while other areas undergo simpler processing. This local quality approach ensures high detection accuracy where needed while reducing overall processing time.
Solution Approach 2:
Before applying detailed pattern recognition algorithms, the patent performs preliminary analysis of subareas using simplified criteria. This preliminary action identifies promising regions that warrant further detailed analysis, reducing the amount of data requiring intensive processing while maintaining detection accuracy for actual patterns.
Data Source
Figure 1
Figure 2
AI summary
The method involves determining whether a pattern correlation quantity determined as a function of difference values is below a predetermined threshold value by an image recognition processor. A determination is made whether the pattern correlation quantity forms a local minimum in an environment of a size of a target object by the image recognition processor. The predetermined pattern is recognized when the pattern correlation quantity is below the predetermined threshold value and the local minimum is present.