Optical signals are detected from by absorbing power from the field. The absorbed power affects the detecting medium by altering its chemical or physical state or by causing charge to flow through it. Photographic film, in which silver halide crystals are converted to elemental silver, is the primary example of chemical detection. Thermal detectors, such as thermopiles and calorimeters, are leading examples of photophysical detectors. In these devices, the incident field raises the temperature and the temperature change is detected using thermometers, thermocouples and pneumatic devices (which detect the thermal expansion of a gas.) Thermal detectors have very slow response times (~seconds) and are used for high power energy measurements. Other detectors, such as biological visual systems, operate through a complex combination of photochemical and electrical effects.
Detectors for optical imaging may be divided between films, which capture optical intensity distributions but do not allow digital transfer and processing, and electro-optic devices, which provide electrical signals proportional to the optical intensity distribution. With the growth of digital processing, transfer and storage devices, electrical detectors are of increasing importance.
Electro-optic detectors convert optical signals into electrical signals. The current density in an electro-optic device is
where 
, 
and 
are the charge density, mobility and polarization density and 
is a bias field. In order to detect radiation, the electromagnetic field must somehow change one or more terms in this equation to modulate the device current. The following approaches may be taken:
devices: In this class, the absorbed light increases the carrier density, thus causing a change in impedance. Two subclasses are:
Photoemissive devices: These are devices based the photoelectric effect. The photo-multiplier tube is the prime example of this class.
Photoemissive Detectors:
The optical field ejects charge from a surface. The charge is accelerated through vacuum and collected on an anode.
Photoconductive devices: These are semiconductor devices in which light excited carriers from the valence band to the conduction band either in a bulk specimen or at junctions.
Photoconductive Detectors:
Conduction band charge is created by the optical field.
devices: This class is loosely referred to as bolometers. These are thermally responsive devices in which optical radiation heats the material, thereby changing the electrical resistance. One subclass operates with 
only and is called a free-electron bolometer because the field effects the carrier temperature directly without affecting the material temperature.
devices: These devices generate an EMF due to a change in the carrier density. They are called photovoltaic devices, an example being the solar cell. One special class operates by displacing electrons and holes in a bulk sample and is popularly called a photon drag detector.
devices: This class is probably the most common approach to electromagnetic wave detection, but it is rare in the optical and infrared region. This is an antenna coupled to a rectifying circuit.
devices: This class relies on pyroelectric changes in the material polarization. This effect is observed in ferroelectric crystals. A ferroelectric is a material with a permanent microscopic scale electric dipole moment. The absorbed radiation heats the sample and generates a current given by:
where T is the temperature, cv is the specific heat and I is the absorbed power/volume.
Direct thermal Detectors:
Optical radiation heats a material and the change in temperature produces a current. This can happen directly, through the pyroelectric effect, or indirectly, as measured by a thermocouple.
A number of issues must be considered in selecting a detector for a particular application. These include:
where 
is the photon energy, 
is the quantum efficiency and e is the electron charge. It is possible to have electron gain, in which case the responsivity is
Gain is achieved, for example, in photomultiplier tubes and avalanche photodiodes.
Gain=(number of dynodes)(electron gain per dynode)
2. Spectral coverage. The responsivity is a function of the radiation wavelength. In semiconductor detectors, the responsivity is shaped by the material bandgap. In photoemissive detectors, the responsivity is shaped by the material work function. Photon counting detectors generally have a more complex spectral dependence than thermal detectors.
3. NEP (noise equivalent power) and D (detectivity). These specify the noise properties of the detection system and determine the minimum detectable power. In imaging detectors, the primary noise sources are counting fluctuations or "shot" noise and dark current or "thermal" noise.
4. Response time.
5. Quantum efficiency
6. Linearity and saturation power.
All of the detectors we have discussed are "square law detectors." The output is not proportional to the optical electric field, rather the output is proportional to the power intensity, which is proportional to the square of the field. All of these detectors also have a finite response time, which is generally much slower than the femtosecond scale period of the optical field. Thus, practical detectors detect the time average square of the field.