GePISCal is a digital image processing project, where the main research is done in the image segmentation, image features extraction, object extraction and classification, and image interpretation areas. It aims the development of an effective and reliable automatic solution for graphic research/surveillance, through the direct confrontation of the graphic elements existing in the images to be compared, without any human intervention in pre- or post-research screening of the images obtained. Through the design and implementation of image analysis algorithms, based on machine and deep learning models and techniques, it is intended to model the human perception of images, and overcome some of the limitations of currently existing Content Based Image Retrieval (CBIR) systems. It is intended that the system resulting from this project be usable in a real context, in which the databases have dimensions in the order of millions of images, and continuous growth.