This article discusses an approach to building systems based on the knowledge with ability to extend their own domain knowledge through obtaining information in the Internet. In that approach proposed to use infinity loop in which the system automatically learns to find good quality documents within a knowledge area represented in Internet using ontology and expert preferences and on next step extend this ontology by extracting knowledge from retrieved documents. Such approach would let to create a system capable of continuously increase their own knowledge by exploring documents in the Internet and solve problems using current state of the knowledge area.

The article briefly describes the methodology and suggests the method for recognizing any elliptic forms objects on the images. This method is universal and can be applied in any intelligent recognition systems, for example, recognition system of the road signs from video camera images. The proposed method has proven itself in solving various practical problems, such as searching for signs in photographs, detecting circles on charts and diagrams, searching for the boundaries of ovals of faces, etc. The main advantage of the method is its extreme ease of implementation and high speed, which makes it possible to use not only on modern stationary computers, but also on mobile devices with low computing power.