Children's Binocular Ranging With Adaptive Stereo Rectification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing binocular ranging methods for children are inaccurate due to complex parameter settings that do not adjust with children's growth, and incorrect camera placement can lead to erroneous measurements, particularly in measuring the distance between a child's eyes and a screen.
Innovation Solution
A method and system for binocular ranging that includes image acquisition, stereo rectification, face detection, and eye gaze direction detection using a binocular camera to accurately determine if a child's eyes are staring at a screen, with a system comprising modules for image acquisition, stereo rectification, face detection, face matching, eye gaze direction detection, and ranging, which uses trained models to identify children and measure distances without preset height or weight adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If binocular ranging is performed using preset parameters for children, then ranging function can be implemented, but measurement precision deteriorates because parameters do not adjust with children's growth
Solution Approach 1:
The patent applies dynamics by making the ranging parameters adaptive rather than fixed. The system automatically adjusts camera intrinsic parameters (focal length, principal point) and extrinsic parameters (position, orientation) based on real-time detection of the child's face position and distance. This dynamic adjustment ensures measurement precision is maintained as the child grows and changes position, resolving the contradiction between having a working ranging function and achieving accurate measurements.
2Measurement precision
If camera placement requirements are made strict to ensure measurement accuracy, then measurement precision improves, but device complexity increases due to complex installation and adjustment
Solution Approach 1:
The patent applies self-service by enabling the system to automatically calibrate and adjust itself without requiring complex manual setup. The camera performs self-calibration by detecting the child's face and using the known face dimensions to compute and adjust its intrinsic and extrinsic parameters automatically. This eliminates the need for complex installation procedures and strict placement requirements, reducing device complexity while maintaining measurement precision.
3Adaptability or versatility
If face recognition model is trained with diverse child face data, then adaptability improves for different children, but loss of time increases during model training
Solution Approach 1:
The patent applies preliminary action by pre-training the face recognition model with diverse child face data during the system development phase. This preliminary training creates a robust model that can recognize various children's faces. During actual operation, the pre-trained model is already available and ready to use, so no additional training time is lost during deployment. The system may perform quick adaptation to individual children, but the bulk of the adaptive capability is established in advance.
Data Source
AI summary
A method for binocular ranging for children is provided in this disclosure, which includes following steps: acquiring an image of a target area and world coordinates of a human body in left and right eyes; correcting the image to obtain a corrected image of the target area; performing face detection on the corrected image of the target area to obtain a face image and an eye image; recognizing the face image and determining whether the face image is a child's face; if yes, performing image correction based on the obtained face image; inputting the eye image into a preset eye recognition model for recognition whether the human eye is staring at a screen; and performing ranging on the eye image based on the world coordinates of the human body to obtain a distance between the screen and the human eye, if the human eye is staring at the screen.