Cell Targeting of Plant Cells Array using Uncalibrated Vision-Based Approach

There is an increasing demand for uncalibrated micromanipulation systems for its capability to perform on-site microscopic applications. An example is the auto targeting of plant cells for applications such as injection or extraction of chloroplasts. In addition, uncalibrated micromanipulation approach will also save the tedious calibration effort that is required when a new tool or specimen is loaded. Hence, this paper proposes a method to perform automated multiple-cell targeting in a plant cell array through an uncalibrated vision-based micromanipulation system. An improved version of watershed segmentation technique coupled with generalized cell sizing algorithm is first proposed to detect the centroid of the plant cells, without the usual need of staining. A self-focusing algorithm is also used to detect and focus either the plant cell array or the micropipette tip, or both. The method is then combined with uncalibrated tool focusing and visual track-servo algorithms to manipulate the micropipette. Experiment has been conducted and the proposed method has been tested on a waterweed. The results illustrates the capability of the uncalibrated detection and targeting of multiple plant cells.