A wavelet-based electrocardiogram (ECG) data compression algorithm is proposed in this paper. The ECG signal is first preprocessed, the discrete wavelet transform (DWT) is then applied to the preprocessed signal. Preprocessing guarantees that the magnitudes of the wavelet coefficients be less than one, and reduces the reconstruction errors near both ends of the compressed signal. The DWT coefficients are divided into three groups, each group is thresholded using a threshold based on a desired energy packing efficiency. A binary significance map is then generated by scanning the wavelet decomposition coefficients and outputting a binary one if the scanned coefficient is significant, and a binary zero if it is insignificant. Compression is achieved by 1) using a variable length code based on run length encoding to compress the significance map and 2) using direct binary representation for representing the significant coefficients. The ability of the coding algorithm to compress ECG signals is investigated, the results were obtained by compressing and decompressing the test signals. The proposed algorithm is compared with direct-based and wavelet-based compression algorithms and showed superior performance. A compression ratio of 24:1 was achieved for MIT-BIH record 117 with a percent root mean square difference as low as 1.08%.

Thermal imaging technology can be used to detect stress levels in humans based on the radiated heat from their face. In this paper, we use thermal imaging to monitor the periorbital region?s thermal variations and test whether it can offer a discriminative signature for detecting deception. We start by presenting an overview on automated deception detection and propose a novel methodology, which we validate experimentally on 492 thermal responses (249 lies and 243 truths) extracted from 25 participants. The novelty of this paper lies in scoring a larger number of questions per subject, emphasizing a within-person approach for learning from data, proposing a framework for validating the decision making process, and correct evaluation of the generalization performance. A k-nearest neighbor classifier was used to classify the thermal responses using different strategies for data representation. We report an 87% ability to predict the lie/truth responses based on a within-person methodology and fivefold cross validation. Our results also show that the between person approach for modeling deception does not generalize very well across the training data.publishersversionPeer reviewe

No abstract

No abstract

No abstract