Adaptive Color Correction for Pill Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional pill identification techniques face challenges in accurately identifying pills under non-ideal lighting conditions, leading to potential misidentification due to variations in color appearance, and existing color correction methods are inadequate for ensuring accurate color representation.
Innovation Solution
A method and system that utilize a controlled surface with known characteristics and real-time image diagnostics to automatically correct image colors, employing an adaptive color correction model that samples pixel values from known areas to predict and correct color differences across the image, while providing visual cues to users for optimal image acquisition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional color correction methods (e.g., histogram equalization) are used, then the processing is simple and fast, but the color correction accuracy is insufficient for pill identification
Solution Approach 1:
The patent applies preliminary action by placing a controlled surface with known color characteristics into the imaging scene before capturing the pill image. This controlled surface serves as a reference that is processed in advance to establish a color correction mapping, which is then applied to the pill image. The controlled surface includes regions with known reflectance properties that are used to pre-calculate correction factors before the actual pill identification occurs.
Solution Approach 2:
The controlled surface acts as an intermediary between the lighting conditions and the pill image. It mediates the color information by providing known reference colors that have been affected by the same lighting conditions as the pill. By comparing the expected colors of the controlled surface against the captured colors, the system creates a correction map that indirectly corrects the pill image colors without directly measuring the pill itself.
2Measurement precision
If ideal lighting conditions are required for accurate color imaging, then color accuracy is improved, but the convenience and ease of operation deteriorates
Solution Approach 1:
The system applies self-service by automatically capturing and processing an image of the controlled surface along with the pill image in a single operation. The controlled surface image is automatically used to generate the color correction map, which is then automatically applied to the pill image. This eliminates the need for manual lighting adjustment or separate calibration steps, allowing the system to correct its own color accuracy issues autonomously.
Solution Approach 2:
The patent changes the parameter of color accuracy from being dependent on external lighting conditions to being dependent on the controlled surface reference. By introducing known color parameters of the controlled surface into the imaging equation, the system transforms the color measurement problem into a relative measurement problem where the correction is based on the ratio between expected and observed colors of the controlled surface, making it invariant to lighting changes.
3Reliability
If color correction is performed without a controlled reference surface, then the imaging process is simpler, but the reliability of pill identification under varying lighting conditions deteriorates
Solution Approach 1:
The controlled surface serves multiple functions simultaneously: it provides color reference for correction, establishes lighting conditions characterization, and acts as a focus and exposure reference. This multi-functionality increases reliability without proportionally increasing complexity, as a single component (the controlled surface) performs multiple correction-related tasks that would otherwise require separate systems.
Data Source
AI summary
A system and method for correcting colors in an image in order to identify pills. The system includes a surface that includes a background upon which pills are positioned and a border of known color adjacent to the background. The system also includes a digital camera and a processor. The processor is used to receive and process the image, use target objects on the surface to identify areas of the image that correspond to the known characteristics and features of the surface, perform image diagnostics to determine if the image can be used, and correct image color based on comparison of identified areas of the image that correspond with the known characteristics and features of the surface. Various graphical displays or cues can assist a user in obtaining a usable image.