Digitization of documents has gained prominence in the recent past for data preserving. Paper documents can be converted to digital form by using various modes of acquisition techniques. In this paper processing of data captured using normal digital camera has been considered. The camera captured document images may contain warped document due to perspective and geometric distortions. Curvature of text lines is a prominent characteristic of such distortions, which can affect the recognition accuracy of Optical Characters for OCR (Optical Character Recognition). To improve the accuracy, rectification of perspective distortions plays an important role. Here, a rectification technique that detects and straightens text lines with curvature in document images captured using normal digital camera is proposed.