engleski

Mulitresolution Approach to Processing Images for Different Applications – Interaction of Lower Processing with Higher Vision

Nowadays, due to rapid advancement in technology, especially development in sensor technology, and computational power, lot of automated systems are developed and incorporated. The purpose of such systems is to make our everyday lives easier. Cameras are among most widely used sensors in automated systems. Fields like medicine and engineering almost cannot be imagined without using cameras. For example, nowadays, surgeries are performed with cameras whether the camera is needed for surgeon to see what is doing or the camera is used to transmit information over the Internet or perhaps the camera is used as a diagnostic tool. Also, surveillance applications are very important and widely used, for example in traffic surveillance or to protect some area or buildings, as well as in the computer vision applications in the automatic manufacture industry and quality control systems. Obviously, due to cameras lot of information’s need to be processed, and accordingly it can be seen that image processing is one extremely growing field. Mainly, the purpose of image processing is to reliably collect and send information’s, such as video stream or images, from the source to the destination. Above all there is need to transmit information’s in real time. Many different techniques and methods are employed in order to process the image information and lot of problem’s have to be resolved when processing it. Among them, illumination variations are one of the problems that have to be resolved. Without resolving it many serious problems could arise concerning reliability of built systems and that could lead to potentially hazardous situations. In order to resolve it some transformation has to be used. This work presents an overview of wavelet approach to solving various problems in image processing and visualization of data. In this work, a modern approach of solving a problem of illumination variations is presented by performing time-frequency analysis method known as wavelet transform. It will be presented how with wavelet pair it is possible to resolve problems of illumination variations in surveillance systems. Also, there will be discussion of what further research is needed in order to improve robustness of mentioned systems.

Wavelet transform; Image processing; Scientific visualization; Augmented/virtual reality application; Visual quality control

nije evidentirano