Abstract
Extraction of hidden text in web images, computer screen images, news, games and e-learning is a very important task in image processing. Compression of digital images leads to poor visual quality of background and text images. Digital images are significantly considered and segmented using DWT into text and background blocks. Huffman coding is used to perform the lossless compression process in the compressed text pixels and the SPIHT algorithm in employed to the compress the background pixels. The result of DWT segmentation shows fringes in the segmented text image. The proposed method uses connected region and edge detection approach which provides a segmented text from digital video stills. The segmented text is converted to binary image using luminance thresholding which leads to fine quality of extracted text.
Highlights
Dwt Using Block-Based SegmentationIf the standard deviation (σ) of high frequency sub-bandvalue is greater than the threshold value, the corresponding image block is considered as text block
The combination of text, graphics and natural pictures is called as compound images
The proposed method can be improved with colour segmentation and deleting false regions.Transform domain techniques are implemented in image, audio,and video steganography to enhance the amount of secret information hidden inthe cover object, maintaining good quality of stego object and extracted secretinformation
Summary
If the standard deviation (σ) of high frequency sub-bandvalue is greater than the threshold value, the corresponding image block is considered as text block. These sub images has four subbands namely low-pass (LL), Diagonal(HH), Vertical (LH and Horizontal (HL). The edges are enhanced in all aspects saying vertical, horizontal and diagonal using sub-bands (LH, HL, HH).The text that are present in vertical, horizontal or diagonal directions are characterised highly by the high contrast edges. Based on this a high clarification text block is obtained by the high-frequency wavelet coefficients. Before sending information of each block to the decoder, encoding is done in the prediction residual by Huffman encoder
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
