1. Research paper for arbitrary oriented team text detection in video images using connected component analysis
Authors : Kiran Dogra
Pages : 137-141
DOI : http://dx.doi.org/10.21172/1.74.018
Keywords : image, pixels Abstract :An end-to-end system for text detection and recognition is important in multiple domains such as content based retrieval systems, video event detection, human computer interaction, autonomous robot or vehicle navigation and vehicle license plate recognition. There are several commercial systems for text recognition in scanned document. However, these systems typically need cropped and binarized text regions to perform well for natural scene text. Text detection in natural scenes is a challenging problem and has gained a lot of attention recently. Such texts present low contrast with background, large variation in font, color, scale and orientation combined with background clutter. Therefore a robust and fast recognition system is desirable. In this work, we have explored the dominant text detection in video frames in which variety of images has been selected which has arbitrary alignment of text as well as varying color and size of the text. In this work, we have used wavelet based laplacian filters as well as laplacian of Gaussian filters to get the arbitrary directed text edges in a frame. After that fuzzy clustering has been applied to get the foreground text edges on which dilation and erosion operations are applied to get the pixels containing whole text. After that MSER technique has been implemented along with connected component analysis to extract the prominent text pixels in the image. Experimental results show that the present technique gives good recall and precision ratio on the tested databases.
Citing this Journal Article :Kiran Dogra, "Research paper for arbitrary oriented team text detection in video images using connected component analysis", Volume 7 Issue 4 - November 2016, 137-141
Click here to Submit Copyright Takedown Notice for this article.