Multi Language Text from an Image to English Translation

Multi Language Text from an Image to English Translation

Urmila Shrawankar (GHRCE, Rashtrasant Tukadoji Maharaj Nagpur University, Nagpur, India) and Achint Kaur (Birla Corporation Limited, Kolkata, India)
Copyright: © 2018 |Pages: 17
DOI: 10.4018/IJTD.2018010102


The article elaborates difficulties faced during text identification from an image. The authors evaluated various algorithms in order to find the solutions. The aim of the research is to increase robustness with respect to different challenges. The system solves the problem of reading multiple texts embedded in an image as well as provides a meaningful translation in English language.
Article Preview

Challenges Discovered In Text Identification

Text in image has different appearance and differs in terms of font size, style, orientation, background, alignment, color. Table 1 explains the challenges of each sample image.

Table 1.
Challenges of the system
Sample ImagesChallenges
Figure 1 Font size- Figure 1 shows how unusual font or multiple fonts creates difficulty in analysing the correct character like the dots namely chandrabindu, Anuswar, visarga
Figure 2 Style- Different shape represents the same character but varies in writing style i.e. variation in curve, slants or modifiers as shown in Figure 2
Figure 3 Background- The text image in Figure 3 contains a background image having different contrast and resolution with respect to the foreground image. The image may contain some pictures embedded in it. In that case it becomes difficult to identify text from the image
Figure 4 Alignment- The text alignment in Figure 4 can be any direction which leads to geometric distortions. Also, the distance between characters may not be uniform
Figure 5 Color- Figure 5 shows the text in the image may contain more than two colors in a single word.
Figure 6 Header line over both languages- Sometimes the shirorekha merges with both the languages as shown in Figure 6, So it becomes difficult to segment a word.
Figure 7 Composite characters- The Devanagari script consists of combination of characters like OM, these characters need to be stored carefully in database or else they would not be recognized by the SVM at the time of mapping is shown in Figure 7
Figure 8 Orientation- Unconnected line segments or pixels identification is challenging as even feature extraction does not analyse it properly and the character needs to store in database separately (Figure 8)
Figure 9 Figure 9, illustrates text written in two colours and follow no sequence i.e. any language letter can occur any Where in the word.
Figure 10 In Figure 10, Image contains multiple text lines. It also contains various symbols.

Complete Article List

Search this Journal:
Open Access Articles
Volume 12: 4 Issues (2021): Forthcoming, Available for Pre-Order
Volume 11: 4 Issues (2020)
Volume 10: 4 Issues (2019)
Volume 9: 4 Issues (2018)
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing