KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Event-based Camera
Neuromorphic camera that captures light intensity changes asynchronously rather than complete images at regular intervals. These sensors generate events only when pixels detect significant brightness variations.
Dynamic Vision Sensor
Specific type of event-based camera that responds to contrast changes in the scene with microsecond latency. The DVS emulates the functioning of the biological retina by producing spikes for light-dark transitions.
Silicon Retina
Integrated circuit that mimics visual processing performed by the biological retina at the pixel level. These devices perform early data compression and motion detection directly at the sensor level.
Neuromorphic Cochlea
Biomimetic auditory sensor that converts sound waves into temporal pulse trains similar to the biological cochlea. These devices perform frequency analysis and audio information compression in real time.
Asynchronous Sensing
Capture paradigm where pixels or sensors operate independently and communicate only when they detect significant changes. This approach eliminates temporal redundancy and significantly reduces energy consumption.
Temporal Contrast
Measure of the variation in light or sound intensity between two successive moments used as an event trigger in neuromorphic sensors. The temporal contrast threshold determines the sensor's sensitivity to changes.
Bio-inspired Sensors
Sensors whose design is directly inspired by detection and processing mechanisms of biological sensory systems. These devices reproduce principles such as adaptation, change detection, and temporal coding.
Neuromorphic Engineering
Discipline that designs artificial systems inspired by the architecture and functioning of the biological nervous system. It encompasses the design of sensors, processors, and algorithms that mimic neuronal computation principles.
Spike-based Vision
Computer vision approach using representations based on temporal spikes rather than frame-based images. This method allows for rapid and efficient processing of dynamic visual information.
Event Stream Processing
Continuous processing of event data streams arriving asynchronously and at high frequency. ESP algorithms are optimized to extract relevant information from temporal sequences of events.
Differential Pixel
Neuromorphic sensor pixel that measures changes in brightness rather than absolute values, generating ON/OFF events based on the direction of variation. These pixels enable efficient compression of static visual information.
Latency-aware Computing
Computing paradigm that explicitly considers the temporal delays between events to optimize performance and responsiveness. This approach is crucial for real-time applications using neuromorphic sensors.