Ultra High Speed Digital Imaging

The aim of image sensors is to transcribe, as faithfully as possible, the image of an illuminated object, or light source, formed by a suitable optical system. In the specific context of ultra-fast digital imaging, the additional constraint is to digitize the image received in the shortest possible time. The priority is therefore to have very "sensitive" electronic sensors, in order to achieve a satisfactory signal-to-noise ratio in short exposure times, and to be able to "refresh" the image as quickly as possible to accommodate the next view on the image sensor. For extremely short exposure times, in the nanosecond range, it is still necessary to place light amplifiers in front of the semiconductor sensor. This often results in a drop in image spatial resolution, directly related to the temporal gain achieved. A high-speed sensor is therefore mainly characterized by its acquisition frequency (or speed) in frames/s (or fr/s). Among the main types of semiconductor used, we will focus on the two main families of optoelectronic sensors in use today: CCD and CMOS sensors, and their variants:

• CCD (Charge Coupled Device), which collects, transfers and converts the electrical charge generated by incident photons;

• CMOS (Complementary Metal Oxide Semiconductor), which collects and converts the charge generated by incident photons at the collection site.

Exceeding the limits of high-speed image acquisition (sensitivity, resolution-frequency compromise and total number of images acquired) is a constant challenge for designers, who must compete with technological tricks to increase the performance achieved. Two main categories of high-speed cameras have emerged over the past 15 years: CMOS cameras and CCD cameras with on-site storage. While the former are purely technological advances in standard CMOS sensors, coupled with high-speed transfer techniques on board a high-bandwidth internal memory, the latter are making constant progress, making it possible to acquire and transfer an image very quickly to a storage site that is geographically very close, in the sense of the electronic chip, to its place of collection (the active image). In this way, hybrid techniques using processes close to CMOS operation are implemented on board CCD technology. The only limit is the speed at which the electrons move around the collection and storage sites.

These cameras must meet a precise set of technical specifications, currently defined by users and manufacturers. In particular, they are described by the European Machine Vision Association, under the name EMVA1288. We'll be looking at a number of points relating to these descriptors, including sensitivity, linearity, modulation transfer function and...