Document image analysis has today become an increasingly important field due to the desire to reduce the amount paper documents and archives. Optical character recognition (OCR) systems and document structure analyzers are the essential tools to accomplish this task. It often appears that the document to be recognized is not correctly placed on the flatbed scanner, especially when it comes from a book or magazine. . The result is a crooked scanned image, which poses a real problem for analysis, understanding, segmentation and character recognition. Straightening the input image is then a crucial step in understanding the document. In this report, we propose a straightening method based on Hough transform and filtering algorithms. In addition, the study of the characteristics of the characters is necessary if we want the most faithful reconstruction of the document. This involves the study of the color of the characters, their average boldness and their skeleton. Additionally, most optical character recognition systems are sensitive to the quality and size of the given characters. The tilt angle correction and binarization steps damage characters, especially for small characters. To maximize the efficiency of character recognition, characters are scaled and smoothed using a pixel art scaling algorithm. This report is made up of 2 sections. Firstly, the different algorithms proposed to estimate the tilt angle of a document and secondly the characteristics of the characters study. In this section, we propose a simple method to estimate the tilt angle of a document based on the combination of the Sobel edge detection filter, a filtering algorithm and the Hough transform. This method was designed for fast detection of the middle of the paper......too many pixels are removed during the filtering step, which will then not leave enough information for the Hough transform. Tolerance of this filtering algorithm depends on the window size and the threshold value. The greater the difference between the window size and the threshold value, the more tolerant the algorithm. 1.2.2 Grayscale image filtering algorithm This grayscale filtering algorithm was designed to do the same type of filtering as the previous algorithm, but on grayscale images. The major difficulty, compared to binary images, lies in the distinction between the objects and the background. In binary images we only deal with true or false values, while in grayscale images the range of values ​​is 0 to 255. For the rest of this paragraph we assume that the text has a color darker than the background..