A trainable gesture recognizer

Improvement of the Recognition Rate by Random Forest

“…In This paper we propose a segmentation algorithm, in which text is easily segmented into Lines and Words using the traditional vertical and horizontal projection [6].…”

“…Teague first introduced the use of Zernike moments to overcome the shortcomings of information redundancy present in the popular geometric moments [6,7]. Zernike moments are a class of orthogonal moments and have been shown effective in terms of image representation.…”

Word shape analysis for a hybrid recognition system

“…Vertical bars are identified using local vertical extrema. Preprocessed ink data points [15] are analysed and local vertical extrema are detected. Each ordered pair of a maximum and a minimum within a single stroke results in a vertical bar (MaJ:n-Min n in Figure 10).…”

“…When a word alternative is produced only by a single recognizer its score remains unaffected. When a word alternative is produced by more than one recognizer, new scores are calculated using the recognition scores provided by individual recognizers: 1,sc2 ) , u > 1 (15) where se is the combination score, SC! and sC2 are the scores provided by individual recognizers and a is the confirmation factor indicating how much the confirmation of results by another recognizer should improve the score.…”

Hidden Markov Model for Gesture Recognition

“…Several methods have been used for gesture recognition: template-matching [2], dictionary lookup [31, statistical matching [4], linguistic matching [5], neural network [6], and ad hoc methods. Some of the methods are suitable for only one type of feature representation, while others are more generally applicable.…”