Please use this identifier to cite or link to this item:
http://cmuir.cmu.ac.th/jspui/handle/6653943832/72180
Title: | ระบบรู้จำอักขระด้วยแสงสำหรับตั๋วเงินประเภทเช็คในกระบวนการเบิกจ่ายค่าสินไหมทนแทนในธุรกิจประกันภัย |
Other Titles: | Optical character recognition for business cheque in insurance claim payment process |
Authors: | ชินวัฒน์ งามสม |
Authors: | ภาสกร พรรณจิตต์ ชินวัฒน์ งามสม |
Issue Date: | Oct-2020 |
Publisher: | เชียงใหม่ : บัณฑิตวิทยาลัย มหาวิทยาลัยเชียงใหม่ |
Abstract: | This independent study aims to develop a data pipeline system that is able to transform a printed business cheque into digital numeric data using the OCR technique. This system was developed specifically to enhance the efficiency of the data input process of the insurance claim payment process. The evaluation of the system is in two folds. The first one is to evaluate the efficiency among different algorithms used in building the OCR system based on accuracy and runtime. The selected algorithms are k-Nearest Neighbors (kNN), Support Vector Machine (SVM), and Gradient Boosting Machine (GBM) respectively. Based on empirical experimental, GBM was found to be the most accurate and it demanded the least runtime among the three algorithms. The second evaluation is to appraise based on the result of evaluation survey from 10 experts who are either the developers or the person in charges of claiming process in the insurance industry. The survey results show that both of the accuracy and speediness of the system developed is outstanding and it could satisfy all the ten experts. Therefore, it can be concluded that the purposed system can increase the capability of data input process of the insurance claim payment process. |
URI: | http://cmuir.cmu.ac.th/jspui/handle/6653943832/72180 |
Appears in Collections: | ENG: Independent Study (IS) |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
610632076 ชินวัฒน์ งามสม.pdf | 1.12 MB | Adobe PDF | View/Open Request a copy |
Items in CMUIR are protected by copyright, with all rights reserved, unless otherwise indicated.