(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No.8, 2010
Extraction of Information from Images using DewrappingTechniques
Khalid Nazim S. A.,Research Scholar,Singhania University,Rajasthan, India.Dr. M.B. Sanjay Pande,Professor and Head,Department of ComputerScience & Engineering,VVIET, Mysore, India.
An image containing textualinformation is called a document image. Thetextual information in document images isuseful in areas like vehicle number platereading, passport reading and cargo containerreading and so on. Thus extracting usefultextual information in the document imageplays an important role in many applications.One of the major challenges in cameradocument analysis is to deal with the wrap andperspective distortions. In spite of theprevalence of dewrapping techniques, there isno standard efficient algorithm for theperformance evaluation that concentrates onvisualization.
Wrapping is a common appearance documentimage before recognition. In order to capturethe document images a mobile camera of 2megapixel resolution is used. A database isdeveloped with variations in background, sizeand colour along with wrapped images, blurredand clean images. This database will beexplored and text extraction from thosedocument images is performed. In case of wrapped images no efficient dewrappingtechniques have been implemented till date.Thus extracting the text from the wrappedimages is done by maintaining a suitabletemplate database. Further, the extracted textfrom the wrapped or other document imageswill be converted into an editable form such asNotepad or MS word document. Theexperimental results were corroborated onvarious objects of database.
Dewrapping, Template Database,Text Extraction.
An image may be defined as a twodimensional function
where x and y arespatial co-ordinates and the amplitude of
atany pair of co-ordinates(x, y) is the intensity orgray level of the image at that point. When x, yand the intensity values of
are all finite, thedigital image is composed of finite number of elements where each has a particular locationand value. These elements are called pictureelements, image elements, pels and pixels.Image processing can be broadlycategorized into two classes. The first categorytakes images as input and gives the images asoutput. The other category takes images asinput and gives the attributes of images asoutput. The entire processing can be listed as:(i).
- It involvesmanipulating an image so that the result ismore suitable than original for processing.(ii).
- It involves improvingthe appearance of an image based onmathematical or probabilistic model of imagedegradation.(iii).
Colour image processing
- Colour can beused as factor or basis for extracting features