SPIE Membership Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:
SPIE Photonics West 2019 | Register Today

SPIE Defense + Commercial Sensing 2019 | Call for Papers

2019 SPIE Optics + Photonics | Call for Papers



Print PageEmail PageView PDF

Optical Design & Engineering

Expert system software for lens design

Algorithms derived from optical design experts' knowledge can be used to generate optimal new lens structures.
2 June 2011, SPIE Newsroom. DOI: 10.1117/2.1201105.003526

Even with the prominence of computer-aided techniques, designing a lens, or optical system, still relies on an optical engineer's judgment based on knowledge and experience. Nevertheless, computer programs can speed up the process by quickly generating multiple starting layouts, which define the number, positions and types of individual lens elements in the system. The engineer can select or generate a suitable design with which to start, and then optimize it using software to get a final form that meets given performance specifications.

We proposed one of the first approaches for artificial intelligence-supported computer-aided design of a photographic lens in 1992.1–5 We have also developed an expert system software called STRUCT3,4 to synthesize new lens starting points. Expert systems, which solve problems using a knowledge base gathered from human experts, are computer programs based on learning algorithms and can be trained in the lens modeling process to find optimal starting structures.

We consider professors who teach optical system design as experts. Teaching stimulates a deep understanding of the modeling procedure and helps to extract its structure, principals and rules. Computer programmers can formalize this expert knowledge, logically fusing rules and conclusions to form a heuristic algorithm, which quickly provides a good enough solution to a problem without guaranteeing that it will be the most accurate.

An important step in the process of developing a computer program that generates new lens structures is to get suitable optical system classifications. We have used several types of technical and general classifications to formalize the design procedure.1–5

There are four types of optical systems, depending on the location of the object and its image. These are: afocal (or telescope), where both object and image are at infinity; photographic, where the object is at infinity and image is at a finite distance; micro-objective, where the object is at a finite distance and image is at infinity; and projection (or relay), where both the object and image are at a finite distance.

According to many publications, selecting a starting point for these four types requires different approaches. Therefore, our expert system is based on four different algorithms, one for each type, the most developed component being the one for photographic lenses.

Figure 1 explains how starting point selection is connected with the rest of the optical system design process. Here, J, W, F, L, Q, S and D stand for aperture, field, focal length, spectral range, quality index, back focal length, and aperture stop position respectively, and are the technical specifications for the design.1 They guide the selection process for starting layouts. For better understanding of lens classification we can present it in two ways: 2D and 3D.

Figure 1. The technical classification of an optical system guides the selection of a starting layout, and is connected to the rest of the design process. OS: Optical system. J: Aperture. W: Field. F: Focal length. L: Spectral range. Q: Quality index. S: Back focal length. D: Aperture stop position.

An example of the technical classification for a fish-eye lens is given in Table 1. Figure 2 shows its 3D class position while Figure 3 shows the starting structure. In the table, 0, 1, and 2 symbolize the complexity of implementation of optical specifications in an optical class. Here, 0 stands for the simplest system and 2 for the most complex, while 1 denotes average complexity.

Table 1. Example of an optical system (OS) technical classification for a fish-eye lens.
Wide-angle lens Fish-eye type
J ApertureW FieldF Focal length, mmL Spectral rangeQ Quality indexS Back focal length, mmD Aperture stop position
OS class1201210

Figure 2. 3D J-W-F class position for a fish-eye lens.

Figure 3. Starting point for a complex fish-eye lens.

To better understand the complexity of optical systems, we introduced the index of complexity R,5 the sum of the complexity values of all six specifications. For the fish-eye lens example, R is the sum of 1, 2, 1, 2, and 1, which is 7. In terms of our classification this lens is called a ‘fast super-wide angle lens working in the visible spectral range, having intermediate image quality, long back focal length and aperture stop located inside the lens.’

From our experience, if R is greater than 5, the optical system is complex and could require a technical solution, which calls for know-how in optical design. An example of a starting point for such a lens is presented in Figure 3. The first element, S(AP(, serves to increase the field angle of view. It has two surfaces: A, which is aplanatic about the marginal ray and P, which is concentric about the principal ray. The basic element B)AP), which forms the optical power of the lens, is made of the same types of surfaces as S(AP(. The fast element C)AP) develops the aperture. Finally, the two correction elements K(PP( and K)FF)—F is a surface concentric about the marginal ray—are intended to correct residual aberrations of the other elements that could cause the image to blur.

The proposed expert system is now used in optical design classes at the National Research University of Information Technologies, Mechanics and Optics, Saint Petersburg, Russia. We plan to further develop the software components for afocal, micro-objective and projection lens starting point selection.

Irina Livshits, Vladimir Vasilev
National Research University of Information Technologies
Mechanics and Optics (NRU ITMO)
Saint Petersburg, Russia

Irina Livshits received her PhD in 1980 from NRU ITMO. She is the head of computer-aided design at the NRU's opto-information and energy saving systems laboratory. Her interests are imaging, non-imaging optics, expert systems, and optical design software. She is the author of 130 articles, 65 patents, and one book.

Vladimir Vasilev received his PhD from Leningrad Polytechnic Institute in 1980. He joined the NRU ITMO in 1983, became professor in 1992 and rector of the University in 1996. In 2004, he became the Chairman of the Consul of all universities of Saint Petersburg. He has more than 150 publications and is also the President of the D.S. Rozhdestvenskyi Optical Society of Russia.

1. Irina Anitropova, Simple method for computer-aided lens design with the elements of artificial intelligence, Proc. SPIE 1780, 1992.
2. Irina Livshits, State-of-the-art in Russian and Former Soviet Union optical design, 2nd Int'l Conf. on Optics-Photonics Design & Fabrication, pp. 285-287, 2000.
3. I. Livshits, A. Salnikov, I. Bronchtein, U. Cho, Database of optical elements for lens CAD, 5th Int'l Conf. on Optics-Photonics Design & Fabrication, pp. 31-32, 2006.
4. I. L. Livshits, I. G. Bronchtein, V. N. Vasiliev, Information technologies in CAD system for lens design, Proc. SPIE 7506, 2009. doi:10.1117/12.837544
5. Irina L. Livshits, Vladimir N. Vasiliev, Optical systems classification as an instrument for getting knowledge from optical expert, 7th Int'l Conf. on Optics-Photonics Design & Fabrication, pp. 13-14, 2010.