Dynamic Food Quality Scoring via Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing food product sorting methods fail to differentiate between defects based on their relative area and intensity, leading to unnecessary rejection and waste, as they treat all defects equally regardless of the product's size and consumer preferences.
Innovation Solution
A method and apparatus that capture images of moving food products, perform image analysis to determine color intensities, and score products based on a percentage of color, allowing for individual quality scoring and group appearance scoring, with the ability to adjust quality thresholds dynamically to prioritize defect rejection and minimize waste.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If existing sorting methods treat all defects equally regardless of relative area and intensity, then defect rejection is simplified, but food product waste increases
Solution Approach 1:
The patent applies parameter changes by transitioning from binary defect detection to multi-parameter quality scoring. The system evaluates multiple parameters including defect area percentage, defect intensity, and overall product appearance score. By changing the detection parameters from simple presence/absence to continuous quality metrics, the system can differentiate between minor and major defects, allowing acceptable products with minor flaws to be retained while rejecting only those with significant defects.
Solution Approach 2:
The patent implements dynamics through adaptive quality thresholds. Instead of using fixed rejection criteria, the system dynamically adjusts quality thresholds based on product type, defect severity, and desired quality levels. The scoring system allows for flexible threshold setting that can accommodate different quality requirements for different product categories, enabling optimized rejection decisions that minimize waste while maintaining quality standards.
2Measurement precision
If image analysis is used to detect defects, then sorting precision is improved, but device complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the image analysis process into distinct functional modules: image capture, preprocessing, defect detection, quality scoring, and rejection decision-making. The system segments the product image into different regions (product area, defect areas, background) and evaluates each region separately. This modular segmentation simplifies the overall system complexity by breaking down the complex image analysis task into manageable, independent processing stages.
Solution Approach 2:
The patent uses an intermediary scoring system that bridges the gap between complex image analysis and simple rejection decisions. Instead of directly translating image data into rejection/acceptance decisions, the system introduces intermediate quality scores that summarize multiple defect parameters into a single evaluative metric. This intermediary scoring layer simplifies the decision-making process while maintaining the precision benefits of comprehensive image analysis.
3Productivity
If static defect thresholds are used for sorting, then sorting speed is maintained, but quality control adaptability decreases
Solution Approach 1:
The patent implements dynamics by replacing static defect thresholds with adaptive quality thresholds. The system can adjust quality thresholds dynamically based on product characteristics, defect types, and desired quality levels. Different threshold configurations can be applied to different product categories or quality requirements without changing the underlying image analysis system, enabling flexible adaptation while maintaining high sorting speeds through efficient threshold-based decision-making.
Data Source
Figure 1A~1B
Figure 2
Figure 3
AI summary
A method for scoring and controlling the quality of dynamic food products transitioning in the processing steps is performed using image analysis. An image of a plurality of moving food products on a conveyor system is captured by on-line vision equipment and image analysis is performed on the image via an algorithm that determines the percentage of pixels having varying intensities of colors and applies predetermined preferences to predict consumer dissatisfaction. The entire group of food products of one or more images is given an overall appearance score and each individual food product is also scored such that each may be ranked from least to more acceptable. The ranked food products can then be ejected in the order of worst to better rank to increase the overall quality score of the entire group.