Databases Reference
In-Depth Information
FIGURE 7.3: Estimated execution time for minimum voting function in the
coordinator node for 10,000 pattern classes as a function of the number of
features.
post-processing mechanisms described in this section do not entail a rigid
framework. Different types of data analysis and feature extraction can be ac-
commodated in this scheme. The DHGN multi-feature scheme is considered to
be commodity application that can be used in different application domains.
7.3 Handwritten Object Classification with Multiple
Features
In this section, we demonstrate the capabilities of the DHGN distributed
scheme as a single classifier for combined multi-feature pattern recognition on
handwritten character objects. A comparative evaluation with previous work
performed by Duin and Tax [88] will be discussed. Note that this work on the
DHGN multi-feature scheme is not intended to showcase an optimal solution
with high accuracy for complex pattern recognition. Rather, this study was
carried out to provide an alternative scalable pattern recognition scheme for
multi-feature patterns.
