Automatic method and system for similar images and image fragments detection basing on image content

Automatic method for similar images and image fragments detection based on image content, characterized in that through a computer system, which control unit contains additional interconnected modules:

• module for input configuration, to setup data storage and input image;
• module for image segmentation;
• module for segments processing;
• module for numeric vector development;
• module for numeric vector storage;
• user module for data representation, are carried out the following stages:
• a) specifying at least one input image;
• b) specifying options of the input image processing;
• c) carrying out the input image processing, according to the options, selected at stage b);
• d) choosing through the segmentation, at least one geometric shape;
• e) normalizing indicated geometric shapes and computing area ratio to the length of the outline Sn/Ln of the normalized geometric shape for creation of the numerical vector Ven, and additionally geometric parameters:
• area Sn, outline length Ln, shortest projection Psn and longest projection Pln on the X and Y axis, and a number of outline angles An of the geometric shape;
• f) creating numeric vector Ven, which includes geometric parameters calculated at stage e), for each geometric shape selected at stage d), wherein the numeric vector Ven expressed by the following formula:
• Ven=[Sn1/Ln Sn2/Ln Sn3/Ln3 . . . Snk/Lnk],
• where k—a number of geometrical shapes (segments) allocated on the image;
• g) saving numeric vector built at stage f) in data storage and in module for numeric vector storage of the control unit;
• h) calculating the difference dV between the numerical vector Ven, built at stage f), and image numerical vector Vdb previously saved in a data storage for all comparative images;
• i) identifying images as similar images, if the dV difference less than the specified boundary value.