Multistage Model Training via Histogram Hot Zone Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multistage process model training methods are inefficient in identifying and addressing errors in manufacturing processes due to the need for extensive training data and redundant image processing, which increases training time and data complexity.
Innovation Solution
The method involves generating gray level histograms for input images from different stages of a manufacturing process, determining change values between these histograms, and identifying a 'hot zone' of significant changes, which is used to focus the model training on specific areas of the images, reducing the amount of data needed and streamlining the training process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If extensive training data and redundant image processing are used, then model training completeness is improved, but training time and data complexity increase
Solution Approach 1:
The patent extracts and processes only the most informative features from images using gray level histograms and change value calculations. By focusing on histogram changes between consecutive images rather than processing entire images, the method extracts essential information while discarding redundant data, thus reducing training time while maintaining model training completeness.
Solution Approach 2:
The patent applies local quality by identifying and focusing on specific regions of interest through hot zone detection. Instead of uniformly processing all image data, the method concentrates computational resources on areas with significant changes (hot zones), thereby reducing overall processing complexity and training time while maintaining effective model training in critical areas.
2Reliability
If extensive training data and redundant image processing are used, then model training completeness is improved, but data complexity increases
Solution Approach 1:
The patent transforms image data into a different parameter space using gray level histograms. By converting spatial image information into histogram distributions and then analyzing changes in these distributions, the method simplifies the data structure while preserving essential information needed for model training, thereby reducing data complexity without compromising training completeness.
Solution Approach 2:
The method extracts only the essential characteristics from images through histogram analysis and change value calculations. By taking out only the relevant features (histogram changes in hot zones) rather than processing complete images, the patent reduces data complexity while maintaining the information necessary for complete model training.
3Area of stationary object
If general image processing is used, then comprehensive coverage is improved, but defect detection precision decreases
Solution Approach 1:
The patent applies local quality by identifying hot zones - specific regions where significant changes occur between consecutive images. By concentrating analysis on these localized areas rather than uniformly processing entire images, the method achieves both comprehensive coverage of relevant changes and high precision in defect detection within those critical regions.
Solution Approach 2:
The patent segments the image processing task by dividing it into distinct steps: histogram generation, change value calculation, threshold comparison, and hot zone identification. This segmentation allows the system to comprehensively analyze images through systematic processing while focusing computational precision on specific defect-related regions rather than dispersing effort uniformly across all areas.
Data Source
AI summary
Techniques for multistage process model training are described herein. Another aspect includes determining a first gray level histogram corresponding to a first input image. Another aspect includes determining a second gray level histogram corresponding to a second input image. Another aspect includes determining a set of change values, each change value corresponding to a change in a respective gray level from the first gray level histogram to the second gray level histogram. Another aspect includes comparing each change value of the set of change values to a threshold. Another aspect includes, based on determining that a first change value of the set of change values is higher than the threshold, adding a first gray level corresponding to the first change value to a hot zone of the second input image. Another aspect includes training a model using the hot zone of the second input image.