Development and assessment of image reconstruction algorithms using a low-cost bench-microscope based on a linear CMOS image sensor

Abstract

We aim at establishing a bench-microscope based on a linear sensor as a versatile research tool for the development and assessment of image reconstruction algorithms. Therefore we built a laboratory prototype of a scanning-stage bright-field microscope. It is epi-illuminated with a white-light source. The detector is a linear CMOS image sensor. A stand-alone sensor readout module has been developed and integrated in this low-cost bench-microscope. Through lateral scanning along one axis it acquires two-dimensional (2D) images of the specimen that suffer from substantial contributions from out-of-focus portions. This haze in the optical slices will be removed or significantly diminished using computational methods. As the output performance of these methods is extremely dependent on the imaging quality of the microscope prototype, two types of measurement to assess it are described. Axial discrimination will be evaluated along with the development of computational methods. Nevertheless a plane reflector was scanned along z-axis to measure intrinsic axial response of this microscope arrangement. Results of overall system resolution and contrast are presented. At last, preliminary results of three-dimensional (3D) image reconstruction using a simple algorithm to find the best-in-focus image through the determination of maximum intensity are presented. Raw bright-field images from wire bond of an integrated circuit (IC) were used.