Genetic Programming Image Recognition with Fitness-Based Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image recognition programs face challenges in achieving high positional recognition accuracy due to large variations in precision depending on the input image, and existing methods struggle to evaluate robustness effectively, leading to potential misidentification and low robustness in image recognition algorithms.
Innovation Solution
An apparatus and method that uses genetic programming to create an image recognizing program by combining partial programs, calculating similarity maps, and evaluating fitness based on the distribution of these maps to select programs with high robustness and precision, thereby improving the accuracy of position detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If genetic programming is used to automatically create image processing programs by combining partial programs, then automation and adaptability are improved, but measurement precision and reliability of positional recognition deteriorate due to large variations in precision depending on input images
Solution Approach 1:
The patent implements feedback mechanisms by calculating fitness values based on similarity map distributions and using this information to guide the genetic programming evolution process. The system evaluates each individual program's performance through similarity calculations and uses this feedback to select and optimize programs, thereby improving positional recognition accuracy while maintaining automation.
Solution Approach 2:
The patent changes parameters by evaluating programs based on similarity map distributions rather than simple accuracy metrics. It transforms the optimization criteria by calculating fitness from distribution characteristics, allowing the system to adapt program parameters to achieve more consistent positional recognition across different input images.
2Ease of operation
If existing image recognition programs are used, then simplicity and ease of operation are maintained, but reliability and robustness deteriorate due to potential misidentification and low robustness under varying conditions
Solution Approach 1:
The patent enables the system to self-optimize by automatically evaluating and selecting image processing programs based on their performance in calculating similarity maps. The genetic programming framework allows the system to autonomously improve its own reliability by evolving programs that demonstrate better robustness, without requiring manual intervention while maintaining ease of use.
Solution Approach 2:
The patent performs preliminary evaluation of candidate programs by calculating their fitness based on similarity map distributions before deployment. This preliminary assessment ensures that only reliable programs with demonstrated robustness are selected, preventing misidentification issues before they occur in actual application.
3Productivity
If fitness evaluation is performed without considering similarity map distribution, then calculation speed and productivity are improved, but measurement precision and reliability deteriorate due to inability to evaluate robustness effectively
Solution Approach 1:
The patent applies partial evaluation by calculating fitness based on distribution characteristics rather than exhaustive testing. It uses representative samples and distribution statistics to assess robustness, achieving a balance between evaluation speed and accuracy by not requiring complete exhaustive analysis while still capturing essential robustness properties.
Data Source
AI summary
An apparatus stores a plurality of partial programs, which are constituent elements of an image recognizing program that detects a position of a template image on an input image. The apparatus creates a plurality of individual programs each being a combination of at least two of the plurality of partial programs, and calculates a similarity map that associates similarity with the template image with each pixel of the input image by using each of the plurality of individual programs. The apparatus calculates fitness for each of the plurality of individual programs, based on a distribution of the similarity map, selects an individual program for which the fitness is equal to or greater than a prescribed threshold, from among the plurality of individual programs, and outputs the selected individual program as the image recognizing program.