Program Generating Apparatus for Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The process of automatically generating image processing programs using genetic programming is time-consuming, especially when a large number of selectable partial programs are available, as it increases the likelihood of taking longer to find an appropriate combination of programs for high-quality image processing.
Innovation Solution
A program generating apparatus that sets selection probabilities for parameter values in partial programs based on their effectiveness, using learning data from input and target images, to select optimal parameter values and combine partial programs efficiently through genetic programming, thereby reducing the time needed to generate an image processing program.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a larger number of selectable partial programs are prepared in advance, then high-quality image processing closer to targeted processing is achieved, but the program generating process takes a long time
Solution Approach 1:
The patent pre-calculates and stores effectiveness information for each parameter value combination before the genetic programming process. This preliminary preparation allows the system to quickly reference pre-evaluated parameter effectiveness during program generation, avoiding time-consuming real-time evaluations while maintaining access to comprehensive parameter options.
Solution Approach 2:
The patent replaces the traditional trial-and-error search mechanism with a probability-based selection mechanism guided by pre-calculated effectiveness information. Instead of mechanically testing numerous program combinations, the system uses statistical methods to guide parameter selection, significantly reducing the time required to find high-quality image processing programs.
2Adaptability or versatility
If a larger number of selectable partial programs are prepared in advance, then more configurable options are available, but it raises the possibility of taking a long time until a combination of appropriate partial programs is found
Solution Approach 1:
The patent incorporates feedback mechanisms where the effectiveness of each parameter value is pre-evaluated and stored. During the genetic programming process, this feedback information guides the selection of parameter values, allowing the system to leverage comprehensive configurability while using effectiveness feedback to prioritize promising parameter combinations and reduce search time.
Solution Approach 2:
The patent changes the approach from exhaustive search to probability-based parameter selection. By pre-calculating effectiveness metrics for different parameter values and using these to guide selection probabilities, the system can explore a wide range of configurable options efficiently, finding appropriate program combinations faster without sacrificing adaptability.
Data Source
AI summary
A setting unit sets a selection probability for each of a plurality of parameter values selectively configurable in a parameter-variable program amongst a plurality of partial programs, based on a relationship between the parameter values and an amount of characteristic obtained from at least an input image included in learning data. When having selected the parameter-variable program as a partial program to be newly assigned to a target position for a mutation in order to evolve an individual by introducing the mutation, a selection processing unit selects, amongst the parameter values, one parameter value according to the selection probabilities respectively associated with the parameter values, and incorporates the parameter-variable program with the selected parameter value set therein into the individual.