Abstract
KEYWORDS:
biomimetics, biologicaly motivated artificial vision, foveal vision
ABSTRACT:
Biological organisms perform routinely a wide area of tasks that either cannot be replicated at all by artificial means (software, electronics or mechanical systems), or can be replicated with much inferior performance and efficiency. Such tasks include obvious ones like intelligence, abstract thinking, pattern recognition in human beings, but also other tasks, more trivial and seemingly of lower level, routinely performed by creatures considered much less intelligent, such as insects. Flying insects, for example, are quit adept at moving in complex, unpredictable and possibly hostile and hazardous environments; animals navigate through various uncontrolled environments with seemingly little effort. It is our belief that simple vision and motion cues inspired from nature can successfully be used for autonomous navigation. Hence, an important idea is to study how biology achieves these tasks, in order to learn from them and be able to re-engineer systems according to this new knowledge. This is not an uncommon philosophy, but an emerging philosophy in the engineering field, known as Biomimetic or Biomorphing Engineering. Basically it's the concept of taking ideas from nature and implementing them in another technology such as engineering, design, computing, etc. Although the engineering described by these terms encompasses all kinds of levels, from the molecular level to that of an ecosystem, the concepts are extremely relevant, given the fact that whole biological systems often outperform artificial systems. The first step towards this is identifying the interesting properties, which often requires the knowledge of a biologist and the skills of an engineer. The paper describes some characteristics of biological vision, common to all kind of animals that worth be taken into consideration when designing an artificial vision system. The paper underlines the use of the foveal vision and of the regions of interest, methods known to significantly reduce the amount of information passed to subsequent processing layers.There has been a growing interest in recent years in video processing techniques that resemble the primate visual system. This technique divides the field of view into a region of maximal resolution, the fovea, surrounded by another region whose resolution decreases towards the periphery. A common implementation for such a system is through log-polar mapping. However a major drawback of this approach is that the parameters of data reduction are fixed, imposed by the physical design. It is the author’s belief that the software implementation of a foveal vision system would be a more flexible approach, enabling such a vision system to be successfully used for any autonomous navigation vehicle.