By Tong Lu, Shivakumara Palaiahnakote, Chew Lim Tan, Wenyin Liu (auth.)
This publication provides a scientific advent to the newest advancements in video textual content detection. commencing with a dialogue of the underlying thought and a short historical past of video textual content detection, the textual content proceeds to hide pre-processing and post-processing recommendations, personality segmentation and popularity, identity of non-English scripts, strategies for multi-modal research and function assessment. The detection of textual content from either traditional video scenes and artificially inserted captions is tested. a number of functions of the know-how also are reviewed, from vehicle plate reputation and street navigation suggestions, to activities research and video ads structures. gains: explains the elemental idea in a succinct demeanour, supplemented with references for extra examining; highlights sensible recommendations to assist the reader comprehend and enhance their very own video textual content detection platforms and purposes; serves as an easy-to-navigate reference, providing the fabric in self-contained chapters.
Read or Download Video Text Detection PDF
Similar video books
The #1 on-the-job tv and video engineering reference. it is a problem to stick in sync with the fast moving global of television and video at the present time. Networking schemes, compression expertise, computing structures, gear, and criteria are all yet some of the issues that appear to alter per month. because the box transitions from analog to hybrid analog/digital to all-digital broadcast networks, stations, video video creation amenities, and success-minded engineers and technicians stay awake to hurry with the one reference monitoring all of the alterations within the box: the "Standard instruction manual of Video and tv Engineering".
“If you have got equipped castles within the air, your paintings don't need to be misplaced; that's the place they need to be. Now placed the principles lower than them. ” - Henry David Thoreau, Walden even if engineering is a learn entrenched firmly in trust of pr- matism, i've got constantly believed its influence needn't be restricted to pr- matism.
The coming of the electronic age has created the necessity to manage to shop, deal with, and digitally use an ever-increasing volume of video and audio fabric. hence, video cataloguing has emerged as a demand of the days. Video Cataloguing: constitution Parsing and content material Extraction explains the best way to successfully practice video constitution research in addition to extract the fundamental semantic contents for video summarization, that's crucial for dealing with large-scale video info.
Additional resources for Video Text Detection
In: Proceedings of the fourteenth international conference on pattern recognition, 1998 11. Kim H-K (1996) Efficient automatic text location method and content-based indexing and structuring of video database. J Vis Commun Image Represent 7(4):336–344 12. Jain AK, Yu BIN (1998) Automatic text location in images and video frames. Pattern Recogn 31(12):2055–2076 13. Zhong Y, Karu K, Jain AK (1995) Locating text in complex color images. Pattern Recogn 28(10):1523–1535 14. Wong EK, Chen M (2003) A new robust algorithm for video text extraction.
Then for a potential text candidate region, the wavelet histogram is computed by quantizing the coefficients and counting the number of the pixels with their coefficients falling in the quantized bins. Next, the histogram is normalized by the value of each bin representing the percentage of the pixels whose quantized coefficient is equal to the bin. Comparing with the histogram of non-text area, generally, the average values of the histogram in text lines are large. As shown in , LH and HL bands are used to calculate the histogram of wavelet coefficients for all the pixels, and the bins at the front and tail of the histogram are found large in both vertical and horizontal bands, while the bins located in the middle parts are relatively small.
40) which can be solved according to the theory of generalized eigenvalue system. Although there exists a major stumbling block that an exact solution to minimize the normalized cut is an NP-complete problem, approximate discrete solutions x can be achieved efficiently from y. Some examples of normalized cut are shown in Fig. 13. Graph-based segmentation  is an efficient approach by performing clustering in the feature space. Such a method works directly on the data points in the feature Fig.