Blink Pattern Communication Using Computer Vision in Low Light
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing assistive technologies for fully body disabled individuals fail to effectively communicate with humans nearby using eye blink patterns, often requiring complex eye movements, are costly, or ineffective in low light conditions.
Innovation Solution
A computer vision-based software system that detects and correlates eye blink patterns to phrases, outputting them through a GUI and audio, using a portable setup with an infrared camera and machine learning algorithms to recognize blinks and convert them into understandable language.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hardware-based eye tracking systems are used to detect blink patterns, then detection precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent uses a standard camera to capture images of the eye, creating a visual copy of the eye structure. This copy is then processed through computer vision algorithms to detect blink patterns, replacing the need for specialized hardware eye trackers while maintaining detection capability
Solution Approach 2:
The patent replaces mechanical/optical hardware-based eye tracking systems with a software-based computer vision system. Instead of using specialized emitters and sensors, the system uses standard camera imaging combined with algorithmic processing to detect eye blinks and communicate with humans nearby
2Adaptability or versatility
If multiple types of eye movements are required for communication, then communication versatility is improved, but ease of operation deteriorates
Solution Approach 1:
The patent extracts and focuses solely on the blink movement from among multiple possible eye movements. By isolating this single, simple action, the system maintains communication versatility through pattern variation (different numbers of blinks) while preserving ease of operation, as blinks are naturally easier to produce than sustained eye movements or facial gestures
Solution Approach 2:
The patent uses periodic blinking patterns (different numbers of blinks within a time window) to encode different communication meanings. This allows versatile communication through a simple periodic action that is easy to perform, eliminating the need for multiple distinct eye movement types
3Device complexity
If reflected light detection is used to measure blinks, then device portability is improved, but measurement precision deteriorates in low light conditions
Solution Approach 1:
The patent utilizes the natural color/reflectivity differences of the eye structure (particularly the contrast between the iris and the white sclera) to detect eye blinks. This approach allows the system to function in various lighting conditions using standard camera sensors, maintaining both portability and adequate detection precision without requiring specialized low-light hardware
4Productivity
If computer-based communication interface is used, then communication speed is improved, but adaptability to human communication deteriorates
Solution Approach 1:
The patent uses the computer system as an intermediary device that translates blink patterns into audible speech or text output. This intermediary function allows the disabled individual to maintain direct eye-to-eye contact and natural blinking patterns while the computer mediates the translation into communicative form that can be understood by humans nearby, preserving both speed and human communication compatibility
Data Source
AI summary
The present invention relates to a software program that is made for allowing people with full body paralysis to communicate through patterns of blinks. The device includes a GUI which displays the camera view facing toward the user, a mesh sequence on the users face to confirm detection of it, and phrases listed on the upper left hand side of the screen with the number of blinks correlating to the phrase. On the backend the software includes a program loop in which all instructions in the program and executed in, a convolutional algorithm for detecting on an FPS basis different parts of the face such as the eyes, ears, nose, and mouth, an algorithm for determining the eye movement sequence eyeblink and an algorithm for counting and matching the number of eyeblinks detected to its phrase and outputting it in the GUI and through audio.


