Marker detection in blurred images for high-precision measurement

In this paper, we propose a method to detect markers in blurred images with high precision. An image blur decreases contrast between a marker and background. This leads to an increase of uncertainty of the marker position. The proposed method reduces the increase of uncertainty of detecting marker position with an increase of image blur. The proposed method estimates degrees of image blur by analysing intensity distribution of the marker contour. The estimated degrees of image blur are used for controlling the window size of edge detection on the marker contour. This adaptive control makes edge detection in blur images precisely. The proposed method of marker detection with the adaptive control minimizes the effect of image blur on precision of the detected marker position. In focused images, fluorescent spherical markers were detected with precision of 1.7 × 10^-3 pixels. In images blurred with 19 pixels, fluorescent spherical markers were detected with precision of 2.5 × 10^-3 pixels with the proposed method, in contrast, with precision of 11 × 10^-3 pixels with a conventional detection method. In images blurred with 27 pixels, fluorescent spherical markers were detected with precision of 4.0 × 10^-3 pixels with the proposed method, in contrast, with precision of 25 × 10^-3 pixels with a conventional detection method. These results show the effectiveness of the proposed method on precise marker detection in blurred images.