Kidger Optics Associates: Michael Kidger Memorial Scholarship: Awardees: Winner 2003

you are here: Kidger Home / Michael Kidger Memorial Scholarship / Awardees / Winner 2003

Michael Kidger Memorial Scholarship: AWARDEE - XUEMIN CHENG

The project of my Ph.D. research is: “Techniques for IntelligentComputer-Aided Design of Optical Systems.”I started my research by helping Prof. Wang to implement a global optimization feature into the GOLD optical design program. This type of global optimization consists of many local optimization runs with or without the escape function using the DLS method.Its efficiency depends on the effectiveness of the local DLS optimization as well as the correct choices of escape function parameters. The implementation of search schemes for the optimum global damping factor and solution vector length accelerated the convergence of DLS optimization and enhanced the performance. The algorithm is successfully implemented in the GOLD program, which becomes an optical software package with a practical global optimization feature for the design of complex optical systems.

In order to further improve the intelligence of the computer-aided design, we established a mechanism for the automatic addition and deletion of optical elements in lens optimization. The criteria used in the scheme employs two lens form parameters, quantifying the symmetry of the optical system and the optical power distribution among the individual lens elements. Experiments show that the scheme is capable of selecting the optimum position for adding an optical surface or element when it is needed to overcome the stagnation of an optimization process, and of observing and deleting a redundant element automatically without impairing the performance of the system.

Xuemin receiving 2003 Award Certificate from Tina Kidger
St Etienne, France, 30 Sep 03

I am also taking part in the project of setting up an expert system for the automatic generation of the initial layout of a zoom lens. It includes the development of optimization algorithms to determine the Gaussian parameters for different zoom mechanisms, and the establishment of a lens group database from which the structure of each lens group in a zoom system can be selected according to the specifications.

Another aspect of optimization that I am planning to study is the way to deal with discrete variables, namely glass materials. I think that optimization algorithms, such as those available using neural network techniques, should be able to handle the glass variables effectively, provided we give the program a correct model of the optical design problem.