Abstarct:   Nowadays, due to the large volume of scanned texts, one of the most important applications of image processing is to restore this information of computer documents and to classify and search them. Classification of texts is the act of tagging or separating a text into one of a predefined category. The purpose of this study is to provide a smart and fast solution to extract Persian lines in scanned texts. The proposed method for Line detection is a two-step algorithm in which Connected Components are extracted using Clustering in the first step and text classification is performed in the second step. Experimental results show that the method presented in this research finds lines baxsed on the of different Fonts size in a scanned map. The collection of images used in this research has been collected from various texts. In selecting this collection, an attempt has been made to use texts with different fonts. Each page of text is captured at a resolution of 300dpi or 200 dpi. The image is taken with a Hp Scanjet 4890 scanner with default settings. The test results show that the proposed system has an accuracy of 99.8%.