Chart Decoder: An Automatic System for Data Extraction from Chart Images.

Abstract

The exponential rise of using visualisation to represent vast amounts of data in the form of chart images is widely regarded as the better approach to represent large amounts of data in the form of chart images around the world, which is convenient way to understand what the data is trying to say, but interpreting these charts for data analysis needs assistance. The goal of this project is to develop an interactive system that extracts and decodes useful information from charts based on the chart image the user enters and get a structured representation of the information. Thus, providing efficiency, less human intervention and data for further analysis. The complex process of automatic chart recognition is divided into multiple tasks in this project, including Chart image classification, Text detection and recognition, Text role classification, Axis analysis, Legend analysis, Plot element detection and classification, and Data extraction. We have trained and evaluated our proposed system on a large collection of synthetic as well as real charts. The results show satisfactory and accurate performance of our implemented system