Lie detection method based on sight tracking

Abstract

The invention discloses a lie detection method based on sight tracking, which mainly utilizes a face detection technology and a pupil center positioning algorithm to track pupils of human eyes and eyemovement tracks of the pupils, and predicts whether the pupils lie or not according to the eye movement tracks of videos. The method comprises the following steps: firstly, carrying out face region detection on each frame of image in a video, and storing facial features to obtain center position parameters of two eyes; carrying out pupil center positioning and coordinate acquisition, and storingthe pupil position of each frame; then, calculating parameters of an eye movement track by using pupil position data of each frame, and obtaining feature vectors of the video by calculating the amplitude and angle parameters of pupil displacement; preprocessing the feature vector data set, and performing noise reduction and normalization processing; inputting the processed data into an SVM (Support Vector Machine), and training a binary classification prediction model capable of judging whether a user lies or not; and finally, obtaining a prediction result by loading the trained prediction model.

Description

technical field [0001] The invention belongs to the technical fields of computer vision, machine learning and image processing, and in particular relates to a lie detection method based on gaze tracking. Background technique [0002] Eye tracking technology is an important part of various perception systems. It has a wide range of applications in many fields such as cognitive science, psychology, human-computer interaction, and medical research. We can rely on the analysis of visual senses to obtain Rich visual information. With the development of computer technology, pattern recognition and image recognition algorithms, the development of optical sensor technology, and the improvement of computer computing power, eye tracking technology has also developed rapidly in recent years. At present, eye tracking systems are mainly divided into wearable eye tracking systems and desktop eye tracking systems. The present invention proposes a desktop eye-tracking system that only nee...

AI Technical Summary

Problems solved by technology

But so far, there are almost no inventions that use machine learning and image processing technology to process the line of sight to achieve the effect of lie detection. More lie detection is based on facial expressions, pulse, ECG measurement, sound and sensor methods.

Images