An empirical enhancement using scale invariant feature transform in text extraction from images
Kirti Kaur Sahota, Lalit Kumar Awasthi, Harsh Kumar Verma · 2017
Techniques for extracting text from videos, images are used to mine useful information. This is the reason that text extraction is a field used in a variety of applications. The sequential process of text extractioninvolves detection of text as first stage. This is followed by text localization and binarization. Finally, the last stage is of text recognition. The current systems are not able to distinguish the non-text key points and text key points accurately. Hence, to overcome this issuea modified algorithm is proposed which uses Scale Invariant Feature Transform (SIFT)keypoints. Comparative analysis of the proposed techniquewith the earlier technique and the proposed method has been found to give better results quantitatively. More accuracy has been achieved in discriminating text key points from non-text key points using the new technique. The sample images have been chosen in a manner that they have varied sizes and formats.