These days, an increasing number of documents are distributed in digital format due to their easy transportation, archiving and hard copy reproduction. This fact increases the number of Valuable Document Counterfeits (VDC) as electronic versions of bills, bank checks and transport tickets. Therefore, several techniques have been proposed to identify VDC such as flashcodes, watermarks and 2D bar-codes. We developed a security pattern to identify the genuineness of a printed document. The legitimate source creates a textured image by writing a visual message using some chosen binary patterns belonging to a database of known patterns. During veri fication, the receiver scans the textured image and applies the detection method in order to verify the genuineness of textured image (i.e. the structure of print and scan (P&S) patterns corresponds to a structure of original pat-terns) as well as to identify the visual message (i.e. the visual message has to be readable after pattern detection). Our contribution consists in a textured pattern-based method that exploits the fact that a counterfeiter does not have access to the original digital document. To detect the patterns, different correlation-based approaches have been tested. We first have modeled the deterioration due to P&S process by using a statistical aggregation process on different P&S realizations. This approach leads to high correlation values, but low detection results. The second approach consisting of correlating the P&S textured image directly with original patterns leads to lower correlation values, but high detection results. Our approach has been tested with limited number of printers and scanners, and with uncoated paper. The proposed recognition method, which aims to increase the significance of central pixels and to decrease the significance of border pixels of each module, improves the final results. This two-step method consists in first separating all modules into two classes (white and black module classes) by minimizing the WMSE values among the original (white and black) modules and P&S modules, then second creating the characterization modules as median or mean images for each class and recognize the white and black modules by minimizing the WMSE values among the characterization modules and P&S modules [ART.6]. Medical image processing is considered as an important topic in the domain of image processing. It is used to help the doctors to improve and speed up the diagnosis process. In particular, computed tomography scanners (CT-Scanner) are used to create cross-sectional medical 3D images of bones. Identifying the scanner model can be very useful for doctors. So we proposed a method for CT-Scanner identification based on the sensor noise analysis. We built the reference noise pattern for each CT-Scanner from its 3D image, then we correlated the tested 3D images with each reference noise pattern in order to identify the corresponding CT-Scanner. A wavelet-based Wiener filter approach has been used to extract the noise. Experimental results were performed on eight 3D images of 100 slices from different CT-Scanners, and our method was able to identify each CT-Scanner separately [CACT.4, CACT.24].