Emotion Recognition System in Images Based On Fuzzy Neural Network and HMM

Abstract

The analysis of human's reflect to images is an important An emotion recognition system based on neuro-task in emotion recognition, which can help us to HMM was proposed to analyze the emotion contained in describe and simulate the human feedback of image. images. This system took an initial step in this direction Emotion recognition in image is interesting but by describing a set ofproposed difficulty metrics based difficult. An important problem of emotion recognition on cognitive principles. Both the emotion semanteme systems is the feature selection. Pitch and energy are extraction and emotion model construction were classical features and are often used in application and considered in this system. They were respectively carried research system. Many other measures (relating to pitch, out by neural networks and HMM. According to the energy, durations, runes, spectral, and intensity) have strong relationship between image notable lines and also been studied mainly for facial emotion recognition.human dynamism sensation, the system usedfuzzy neural In our approach, emotional semanteme extraction and network to establish the mapping and obtained the image emotion model construction are combined to establish emotion semanteme sequence. Then the duple hidden the emotion-analysis system. We not only accomplish the markov model (HMM) was employed to simulate human features selection but also treat the continuous features as emotion transition and finally confirmed different a series human brain simulating signals. It is known that emotion models. The system also considered some outer one image consists of many lines. What the system influences to make the system rules be refined in realistic mainly do is the image constitution processing. Dynamic conditions. The experiment shows at least one emotion textures are sequences of textures that can exhibit certain from an image can be recognized. The results illustrate properties in time, see in [6]. We also concluded some the capability of the developing image recognition methods in dynamic textures extraction. system. This system firstly analyzes the emotion information in image, and then extracts the corresponding textures. After some transform, the output is emotion semanteme Proc. 5th IEEE Int. Conf. on Cognitive Informatics (ICCI'06)

General and Interval Type-2 Fuzzy Face-Space Approach to Emotion Recognition

“…Most commonly used techniques for feature selection in the emotion recognition problem include principal component analysis (PCA) [59], independent component analysis [60], rough sets [42], [61], Gabor filter [62], and Fourier descriptors [25]. Among the popularly used techniques for emotion classification, neural net-based mapping [3], [4], [18], fuzzy relational approach [14], linear discriminate analysis [60], support vector machine (SVM) [8], and hidden Markov model [59], [62] need special mention. A brief overview of the existing research on emotion recognition is given next.…”

Emotion-based image retrieval—An artificial neural network approach

“…In the paper [5] surprise has been added to the above set. Authors of the paper [8] removed disgust from the set, but added neutral emotion and hate.…”

Content-based mood classification for photos and music

“…The gradient information is estimated by different features e.g. a Haar Wavelet Transformation [25], the Sum Of Gradients for the sharpness [22], a Hough Transformation [26] or Canny Edge Detectors together with Wavelet Coefficients [5]. Additionally, Yoo uses a granularity of homogeneous regions as Texture description.…”