Dynamic Vision Sensor（DVS，Dynamic Vision Sensor） is a new type of vision sensor. Compared with traditional frame - based cameras, it can capture changes in the scene with extremely high temporal resolution. Traditional cameras obtain information by periodically capturing images of the entire scene, while DVS operates based on a pixel - level asynchronous response mechanism. Each pixel independently detects changes in brightness. When a pixel detects that the light intensity exceeds a preset threshold, it immediately generates an event, recording the timestamp of the change, the position, and the direction of the change (increase or decrease in brightness). This mode of operation gives DVS significant advantages in handling high - speed motion and high - dynamic - range scenes.

Characteristics Of Dynamic Vision Sensors

High Temporal Resolution ：Since events are generated asynchronously, a temporal resolution on the order of microseconds can be achieved.
Low Latency ：The DVS responds to environmental changes almost in real - time with extremely low latency.
Low Data Rate ：Only the information about the changes is transmitted, which greatly reduces the amount of data that needs to be processed.
High Dynamic Range ：It can adapt to lighting conditions ranging from extremely dim to extremely bright, with a dynamic range of over 120 dB.
Energy Efficiency ：Since it doesn't continuously scan the entire scene but only sends data when there are changes, it consumes relatively low power.

Data Formats

The data generated by a DVS usually exists in the form of an event stream. Each event contains the following key pieces of information:

Position ：That is, the pixel coordinates where the event occurs, usually represented in the form of (x, y).
Polarity ：It indicates the change in brightness. Usually, 1 is used to represent an increase in brightness, and 0 represents a decrease in brightness.
Timestamp ：It represents the exact time when the event occurs, measured in microseconds.

For example, a simple event may be represented as follows:

x: uint16_t
y: uint16_t
polarity: uint8_t // Negative: 0, Positive: 1
timestamp: uint64_t

For example, an event (123, 456, 1, 1235624) indicates that at the given time point, the pixel located at the position (123, 456) has detected an increase in brightness.

The next section.

Tutorial on Using Dynamic Vision Cameras