Pill Counting via Cascading Segmentation of Overlapping Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object counting methods, particularly for pills, often require extensive training on specific shapes and colors and are not suitable for translucent objects, leading to inefficiencies and potential errors.
Innovation Solution
A cascading segmentation method involving background subtraction, morphological operations, and iterative watershed processes with varying sensitivity parameters to accurately count objects, including shape and color verification to ensure reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional object counting methods are used with specific shape and color pattern recognition, then the system can count objects of known types, but it requires extensive training and cannot handle translucent objects or variations in shape and color
Solution Approach 1:
The patent applies segmentation by dividing the image processing task into multiple stages: initial segmentation to identify candidate regions, followed by iterative local segmentation to separate overlapping objects. This multi-stage segmentation approach eliminates the need for extensive training on specific object characteristics while accurately handling various object types including translucent pills.
2Measurement precision
If manual counting is performed, then accuracy can be maintained, but the process is time consuming and repetitive
Solution Approach 1:
The patent replaces manual mechanical counting with an automated image processing system that uses segmentation algorithms. The system automatically identifies, segments, and counts objects in the image, achieving both high accuracy comparable to manual counting and significantly improved productivity by eliminating repetitive manual labor.
3Device complexity
If a single segmentation parameter is used, then the processing is simple, but overlapping objects cannot be properly separated
Solution Approach 1:
The patent implements dynamic parameter adjustment by using different segmentation parameters at different processing stages. The initial segmentation uses one parameter set to identify candidate regions, while subsequent iterative local segmentation uses adjusted parameters to separate overlapping objects. This dynamic adaptation improves separation accuracy without requiring overly complex preprocessing.
4Measurement precision
If extensive training on object patterns is performed, then recognition accuracy improves for specific objects, but the system cannot generalize to new object types
Solution Approach 1:
The patent creates a universal segmentation system that does not require training on specific object patterns. The segmentation algorithms work generically on any objects in the image by analyzing intensity variations and spatial relationships, making the system adaptable to new object types including translucent pills without requiring retraining or adjustment for specific object characteristics.
Data Source
AI summary
A method, apparatus, non-transitory computer readable medium, apparatus, and system for counting objects includes first obtaining a digital image. Then, embodiments segment the digital image using a first segmentation parameter to obtain a first region and a second region. Embodiments then segment the second region based on a size of the second region using a second segmentation parameter different from the first segmentation parameter to obtain a third region. Lastly, embodiments generate a count of objects in the digital image by incrementing based on the first region and the third region.