Accuracy improvement of centroid coordinates and particle identification in particle tracking technique

We have applied an algorithm to improve the accuracy of particle identification and centroid coordinates for each particle image in particle tracking technique. The algorithm introduced two techniques; 1) cutting off by each threshold at the peak in the pixel intensity distribution for each image of local area around the particle, and 2) calculation of the centroid based on pixel intensities in the original image of the particle instead of binarized data. The former properly cuts the noise in the background for each particle which has large variety in level particle by particle due to fluctuating illuminations and out-of-focus particles in the image, and the latter avoids the loss of accuracy by the commonly used binarization. We have demonstrated that the algorithm significantly improves the accuracy in deter mination of centroid coordinates and the correctness in particle identification. We have also validated the advantage of the algorithm in accuracy by applying the algorithm to a sequence of confocal microscopy images of diffusing particles in a polysaccharide solution. This algorithm will be signifi cantly useful in particle tracking technique for biological systems, especially for fluorescence microscopy observa-tions with considerable obstructive stray fluorescent signals. L org,j I sm,j ( m , n ) pixel intensity indexed by ( m , n ) on the x and y axis of L sm,j m = 1, ..., d s pixel row index of L sm,j n = 1, ..., d s pixel column index of L sm,j I T,j calculated cut off intensity of L sm,j L cut,j cut off image of L sm,j I cut,j ( m , n ) pixel intensity indexed by ( m , n ) on the x and y axis of L cut,j C w,j weighted centroid coordinate of L cut,j