A Fractal Approach to Characterize Emotions in Audio and Visual Domain: A Study on Cross-Modal Interaction

In this paper, we try to classify the emotional cues of sound and visual stimuli solely from their source characteristics, i.e., 1D time series generated audio signals two-dimensional matrix pixels affective picture stimulus. The sample data consists six 15 s each pictures, which three belonged positive negative valence, respectively. Detrended Fluctuation Analysis (DFA) has been used calculate long-range temporal correlations or Hurst exponent corresponding signals. 2D analogue DFA technique applied on array pictures contrast emotions. We obtain a single unique scaling signal exponents red/green/blue (RGB) component in images. Cross-correlation (DCCA) (both 2D) degree nonlinear correlation present between clips. To assess proportion cross-modal appraisal, Pearson coefficient was calculated using two modalities. results findings have corroborated with human response study based Likert scale ratings. conclude, propose novel algorithm arousal can be classified scenario only while also attempting them.