Lie detection using acceleration plethysmography signal

• Main Authors: Sidek, Khairul Azami, Shariff, Nur Amirah, Alam, Md Nasibul, Hashim, Nuradilah, Ismail, Ahmad Fadzil
• Format: Article
• Language: English; English
• Published: Faculty of Electronic and Computer Engineering (FKEKK), Universiti Teknikal Malaysia Melaka (UTeM) 2018
• Subjects: TK7885 Computer engineering
• Online Access: http://irep.iium.edu.my/63337/7/63337%20Lie%20detection%20using%20acceleration%20plethysmography%20signal%20SCOPUS.pdf; http://irep.iium.edu.my/63337/13/63337_Lie%20detection%20using%20acceleration%20plethysmography%20signal_article.pdf; http://irep.iium.edu.my/63337/; http://journal.utem.edu.my/index.php/jtec/article/view/3654/2540
• Institution: Universiti Islam Antarabangsa Malaysia

Crime records are reported to show an increasing pattern over the years and it is heart breaking since crimes cases will affect a lot of people especially the victims since they have to face severe loss. Interrogations in crime cases are fundamental since this is the element that will determine the status of the crime and the perpetrator. In order to prevent from any deceptions by the criminals, lie detector might be an invention that will be a great help to separate truth and lies. Due to this, the study proposed a lie detector technique using APG signal. APG signal is the second derivative of PPG that can be obtained by placing detector at the fingertip. Literature reviews on related topics were conducted to gather more information regarding deception detection. In order to realise our objective, the proposed methodology is constructed with data collection as the first step. The data were collected form 10 subjects in form of PPG signals. The next step is signal transformation where PPG signals are converted into APG waveforms and the transformed signals will then undergo pre-processing to eliminate noise. Both techniques use MATLAB as the platform to obtain the output. The following step is feature extraction where the filtered signals undergo segmentation to point out the important information to be used in the next stage. The last step is classification where the extracted data is analysed to perform a conclusion whether the subject is lying or telling the truth. This process involves analysing 3 characteristics of the signals which are the Peak to Peak Interval (PPI), Peak Height Difference and Cardioid graph. Results from the experimentation indicates that PPI is not suitable as a mean to differentiate deception and truth as the difference between these two signals are trivial. Peak Height Difference and Cardioid graph are more suitable to detect lies in both PPG and APG signals since there are significant different in PPG and APG waves when subject are telling lies as compared to telling the truth.