Positioning is one of the most important
operations in color sorting systems for falling grain which
directly affects the performance of these systems. Linescan
Charge Coupled Device (CCD) image sensor is
widely used in the color sorting systems, however, the
processing algorithms are much complex. In this paper, a
complete set of algorithms is presented for the above
systems. First, the Line-Scan Merging (LSM) algorithm, a
novel connected component labeling algorithm that can be
applied on line scan CCD imaging system, is proposed.
The end points of objects will be given by LSM. Second,
the direct least squares ellipse-fitting (direct ellipse-fitting
for short) algorithm is applied to position the single grain
with the end points of it from the result of LSM. Finally,
the direct ellipse-fitting algorithm is modified to position
the connected grains to solve the problem of difficulty in
segmenting the connected objects and accurately
positioning them. The experimental results show that the
proposed algorithms can be successful in positioning the
falling grain.