I discussed a PhD thesis at the College of Engineering at the University of Basra (Designing a system that works in real time that supports prosthetics)

I dealt with the thesis submitted by the student Hanadi Abbas Jaber

Suggesting three types of spatial features based on HOG histogram-oriented gradient algorithm and intensity features represented by H, HI and AIH features. H features correspond to extracting HOG features from HD-sEMG map

The HI features are also obtained by combining the H features with the scalar density feature computed from the map, and finally the hybrid AIH features were produced by combining the H features and the AI ​​density feature matrix obtained from the segmented maps.

Three sub-databases were proposed for evaluation and the proposed feature sets were compared with TD time-domain features and a set of intensity features and ICG center of gravity to show the strength of the proposed features

The thesis aims to propose a set of powerful features to combine with adaptive learning techniques and with the use of HD-sEMG technology in order to improve the performance of the electromuscular system

The thesis concluded that the classifier's performance is superior in terms of accuracy and efficiency, based on the AIH features from other feature sets (ie the advantage of ICG, TD, HI, H).