000			02699nam a22001577a 4500
082			_a005.1,TEH
100			_aTehsin, Samabia _9121692
245			_aExtraction of Caption and Animated Text from Videos / Images / _c Samabia Tehsin.
260			_aRawalpindi, _bMCS (NUST), _c2014
300			_ax,108 p,
505			_aTextual information embedded in multimedia can provide a vital tool for indexing and retrieval. Text extraction process has a lot of inherent problems due to the variation in font sizes, color. Backgrounds and resolution Text detection, localization and tracking are the most challenging Phases of the text extraction process whereas text extraction results are highly dependent upon These phases this dissertation focuses on the text deflection, localization and tracking because of' Their very fundamental importance a bio-inspired text detection, localization and tracking is Developed and presented in the dissertation. Anthropocentric approach of text detection is Studied and is mathematically modeled to design a text extraction process. A novel text segmentation method is proposed covering huge range of text scales, colors and font Styles Segmentation procedure consists of adopted K-means clustering and a fuzzy based Perceptual merging process Two effectual .feature vectors ore introduced .for the classification of the text and non-text Objects First. Feature vector is based upon the human text deflection system and is mathematically represented by the Radon Transform of ‘the text candidate objects. Second. Feature vector is derived from the detailed geometrical analysis of the text contents Union of two Feature vector is used for the classification objects and non-text objects using Support vector machine (SVM) Fuzzy based text tracking mechanism is also introduced in the research that can handle static as Well as dynamic text appearing in videos. The dynamic: text includes the simple animations like Vertical and horizontal scrolling as well as' the complex ones like random movement Scale Change and zoom in/out. Text detection and localization results are evaluated on three publicly available datasets namely ICDAR 201 I, ICDAR 2013 and lPC-Artificial text. Moreover, results are compared with state of the art techniques. Comparison demonstrates the superiority of the presented research. Text tracking dataset is also developed and proposed tracking algorithm is tested on the dataset that demonstrates the applicability of the proposed tracking technique.
650			_aPhD Computer Software Engineering Thesis _9132801
651			_aPhD CSE Thesis _9132802
700			_aSupervised by Dr. Asif Masood _9132796
942			_2ddc _cTHE
999			_c608489 _d608489