Perceived Image Processing for Gaze Tracking Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems struggle to accurately extract low-level features from images contaminated with noise and extraneous features, leading to poor pattern recognition performance.
Innovation Solution
The system employs a Perceived Image Processing (PIP) method that generates a perceived image based on model parameters, compares it with a real image, and iteratively adjusts these parameters to minimize the difference between the two, effectively distinguishing object features from noise using a parametric model and error minimization techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional pattern recognition methods are used to extract low-level features from images, then the processing is simple and fast, but the accuracy deteriorates when images contain noise and extraneous features
Solution Approach 1:
The patent introduces a perceived image as an intermediary representation that mediates between the real noisy image and the pattern recognition process. The perceived image is generated by a perceived image generator that models the expected appearance of the target object, serving as a filter that separates relevant features from noise before pattern recognition occurs
Solution Approach 2:
The system performs preliminary action by generating a perceived image before the actual pattern recognition takes place. This perceived image is created based on a model of the target object and serves as a pre-processed representation that highlights relevant features while suppressing noise, preparing the data for more effective pattern recognition
2Reliability
If the image processing system uses complex models to handle noisy images, then the pattern recognition accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent creates a perceived image that is a simplified copy or representation of the real image, generated based on a model of the target object. This perceived image contains only the essential features needed for pattern recognition, eliminating the need to process all the complexity and noise in the original image while maintaining recognition accuracy
Solution Approach 2:
The system extracts only the relevant features from the real image by generating a perceived image based on a target object model. This extraction process removes extraneous features and noise, keeping only the essential characteristics needed for pattern recognition, thereby reducing computational complexity while maintaining reliability
Data Source
AI summary
An embodiment of the present invention provide a system for measuring and modifying at least one model parameter of an object of an image in order to distinguish the object from noise in the image includes a perceived image generator, an image-match function, and a parameter adjustment function. The perceived image generator produces a first perceived image of the object based on the at least one model parameter. The image-match function compares the first perceived image with a real image of the object. The parameter adjustment function adjusts the at least one model parameter so that the perceived image generator produces a second perceived image of the object that more closely matches the real image than the first perceived image.