A novel detection algorithm of microspheres based on Hough transform and ellipse fitting

Microspheres are used in many applications but their detection is often impeded due to susceptibility to disturbance and low accuracy. An autonomous algorithm was proposed based on Hough transform and least squares ellipse fitting to achieve coarse to fine detection of microspheres. The algorithm is robust and accurate in detection of microspheres' center positons, radii, roundness, deficiency and adhesion, which facilitates the monitoring and manipulation of microspheres. The experiments are conducted to verify the performance of the algorithm. The results demonstrate a nearly 100% successful detection rate and less than 10 μm absolute deviation of radius. The average detection speed is about 100 microspheres per minute.

I. Introduction

Microspheres are used in many applications such as optical tweezers[1], substrates for medical essays[2], photonics[3], polymer science and templates for auto-assembly. While it is necessary to locate microspheres and then measure some parameters through video images to achieve autonomous monitoring and manipulation, this task is often interfered by noises in the environment, uneven illumination and so on.