Improvement processes in postal item classification and sorting systems, mail sorting, machine learning, rough sets

The article presents an adaptive machine learning method for image recognition and processing in automated postal item handling processes. Due to the automatic address recognition process, the most critical segment of an automated postal handling line is an integrated reading machine with video coding. In the reading machine, character recognition occurs via an optical module, where the postal item address is read automatically. The low efficiency of this module increases costs and reduces the throughput of the entire postal handling system. A method based on feature extraction of characters using rough set analysis is proposed. For each class of recognized objects, decision rules and distributions of individual parameters are created, forming a reference database. Meanwhile, data from the video coding process allows for the updating of the reference database for individual character classes.THE PROCESSES OF MACHINE LEARNING FOR CLASSIFICATION AND MAILSORTING SYSTEMS
Abstract
In this article we presented the method of adaptive learning for image processing and recognition for the postal applications. Due to the automatic postal address recognition, the most important element of the automatic sorting and processing post mails is integrated optical character recognition module with videocoding system. This solution can read the addresses data using optical character recognition module. Low efficiency of this module raises costs and reduces the throughput of the mail sorting system. The proposed method based on the character features derived from the rough sets analysis. For each classes of recognized objects (characters) are created the decision rules and distributions of the parameters that form the database. Whereas data from the videocoding process allow to update the database patterns of particular classes of characters. (…)

The article contains 24,065 characters.

Source: Logistyka Magazine

Improvement processes in postal item classification and sorting systems, mail sorting, machine learning, rough sets – PDF file for download