The use of recurrent neural networks for skeletonization and thinning of binary images is investigated. The networks are trained to learn a deletion rule and they iteratively delete object pixels until only the skeleton remains. Recurrent neural network architectures that implement a variety of thinning algorithms, such as the Rosenfeld-Kak algorithm and the Wang-Zhang (WZ) algorithm, are presented. A modified WZ algorithm which produces skeletons that are intuitively more pleasing is introduced.

Emotion recognition of violin music is proposed based on strings music theory, where the emotional state of violin music is expressed by Affinity-Pleasure-Arousal emotion space. Besides the music features from audio processing, three features (i.e., left-hand feature, right-hand feature, and dynamics) with regard to both composition and performance of violin music, are extracted to improve the emotion recognition of violin music. To demonstrate the validity of this proposal, a dataset composing of 120 pieces of authorperformed violin music with six primary emotion categories is established, by which the experimental results of emotion recognition using Support Vector Regression report overall recognition accuracy of 86.67%. The proposal could be an integral part for analyzing the communication atmosphere with background music, or be used by a music recommendation system for various occasions.