1 Noldus Information Technology b.v., Wageningen, The Netherlands
2 Rudolf Magnus Institute for Neurosciences, Utrecht University, Utrecht, The Netherlands
During the past years, the interest in the use of techniques based on video tracking and digital imaging for quantification of behavioral patterns has increased considerably. Also, the capabilities of such systems have increased over time. Early systems were only able to track a single animal under very stringent light conditions, whereas nowadays multiple animals can be tracked simultaneously against a variety of complex backgrounds. In most cases, quantification of behaviors with a video tracking system is performed by analysis of locomotory patterns: a large number of parameters has been developed that can be used to measure the behavior of a single animal or to quantify social interaction between multiple animals.
At Noldus Information Technology, the EthoVision video tracking and behavior recognition system is used as a basis for innovative research on behavior recognition. Since its introduction, EthoVision has mainly been used as an automatic system to measure changes in behavior of animals using continuous path-related parameters. Examples are the use of the total distance moved during a trial as a measure for exploratory behavior or activity and the use of the latency to entry in a predefined target zone as a parameter for search efficiency.
During recent years two different approaches have been studied. The first approach uses the classical method of analysis of locomotory patterns. Complex behavioral patterns are characterized using statistical analysis methods on continuous parameters (object position, velocity, etc.). Using this approach, a series of new parameters is being developed that allow a more detailed measurement of behaviors that occur during a trial. The second approach is based on shape analysis. By analyzing the shape of the animal, behaviors that are not directly related to the locomotion of the animal can be measured. Examples are the detection of rearing based on the surface area and roundness of the animal and recognition of specific body parts (tail and nose of the rat) for accurate detection of social interaction.
The overall objective of this research is the development of an automated system that recognizes animal behavior the way a human observer does.