Polarimetric imaging: principles and applications

Optical polarimetry, which provides additional information to intensity measurement alone, is based on principles that have been well known since Stokes' work at the end of the 19th century. Interest in this type of measurement has been revived in recent decades by the arrival on the market of high-performance optical devices based on liquid crystals, which have made it possible to build low-cost polarimeters that can be combined with CMOS sensors for fast, imaging measurements.

Polarimetric imaging by Stokes and Mueller is one of several possible approaches. The latter provides the complete polarimetric response of a sample in linear optics. It gives access to the birefringence, dichroism and depolarization of the medium under test. For some applications, Stokes imaging is sufficient, as the information we seek is carried by the polarimetric characteristics of the wave that has interacted with the medium.

This type of imaging is used in a variety of fields related to healthcare and manufacturing. Applications include geometry or surface quality control, detection of objects in a scene, and characterization of biological tissue properties for early detection of certain pathologies.

After a few theoretical reminders to define the concepts and quantities used in this article, the principle of polarimetric imaging measurement is presented. The main types of imaging polarimeters are described. Device calibration and measurement noise problems are then discussed. The third section presents a number of applications in healthcare and manufacturing: characterization of biological tissue properties, 2D/3D defect detection, 3D reconstruction, materials analysis and 3D digitization. Finally, applications linked to digital holography, robotics or remote measurement are also presented.

As the measurements produced are images, it is of course possible to apply most image processing techniques to enhance, filter or extract features in these images. Adaptations of these processing techniques to the specificities of polarimetric images (vector nature of the data at each point, physical constraints on possible values, etc.) are also proposed in the literature. However, these aspects of polarimetric imaging are not covered in this article.