Gaze-Based Image Acquisition Controller for Power-Efficient Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing algorithms for digital cameras, such as SURF, SIFT, and ORB, consume significant computing and power resources, especially when applied to large image data, creating a Catch-22 problem where reducing data processing and transmission is desired but selecting objects to limit data is challenging.
Innovation Solution
An image acquisition controller that uses eye tracking to pre-select areas of interest based on a user's gaze, allowing for reduced image capture and transmission by adjusting the sensor area and frame rate based on gaze position and duration, thereby focusing processing on specific regions of interest.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image processing algorithms (SURF, SIFT, ORB) are applied to large image data for object identification, then object recognition accuracy is improved, but computing power consumption and power resources increase significantly
Solution Approach 1:
The patent divides the image acquisition process into segments: full image capture at low resolution, gaze-based region identification, and selective high-resolution capture of only the regions of interest. This segmentation allows object recognition to be performed on smaller data subsets, reducing computational power consumption while maintaining recognition accuracy for relevant objects.
Solution Approach 2:
The patent applies different quality levels to different regions of the image based on user gaze. Regions where the user is looking receive high-resolution capture and processing, while other regions use lower resolution. This local quality differentiation reduces overall computing resources needed while ensuring accurate object recognition in the user's field of interest.
2Use of energy by moving object
If the amount of image data to be processed and transmitted is reduced, then power consumption and transmission resources are improved, but the ability to identify objects for function control deteriorates
Solution Approach 1:
The patent performs preliminary gaze tracking to identify regions of interest before full image processing. By pre-selecting which areas contain objects of interest based on user eye movements, the system prepares a targeted subset of image data that requires less processing and transmission power while still containing all necessary objects for function control.
Solution Approach 2:
The patent introduces gaze tracking data as an intermediary that bridges between full image capture and selective processing. The gaze information acts as a mediator to identify which regions contain objects of interest, enabling the system to transmit and process only the necessary portions of the image while maintaining complete object identification capability.
3Quantity of substance
If eye tracking is used to pre-select areas of interest and reduce image capture, then data transmission and processing requirements are reduced, but device complexity increases
Solution Approach 1:
The patent makes the image sensor serve multiple functions: it captures both the user's gaze position and the scene image data using the same sensor component. This multi-functionality reduces the need for separate specialized sensors and processing systems, thereby reducing overall device complexity while still enabling gaze-based data reduction.
Data Source
AI summary
There is provided an image acquisition controller for acquiring an adapted image. The image acquisition controller comprises an interface towards a first image sensor, the first image sensor being arranged to acquire an image, and an interface towards a gaze tracking unit, the gaze tracking unit comprising a second image sensor for registering a gaze position on a scene associated with the image acquired by the first image sensor. The image acquisition controller is arranged to periodically receive information over the interface towards the gaze tracking unit about at least a position on the scene associated with the image provided from the first image sensor where the scene represents a larger part of an image registered by the first image sensor than intended to be acquired as the adapted image. The image acquisition controller is arranged to provide information over the interface towards the first image sensor about a position on the first image sensor, a size of a sensor area associated with the position on the first image sensor and a frame rate for image acquisition within said area. The image acquisition controller is arranged to receive image data over the interface towards the first image sensor wherein the image data at least comprises image data with the size, the position on the first image sensor, and the frame rate for image acquisition within said area such that the received image data is stored. An image acquisition system or device comprising the controller, a communication device with corresponding image acquisition capabilities, a method of image acquisition, and a computer program for implementing the method are also disclosed.