As the aim is to monitor and control the color changing process
of a roasting chicken in the Combi Oven, the extraction of the precise
chicken area has to be solved via segmentation. Generally,
extracting an object from a given image is an easy task as it is an
object edge detection problem which has been investigated extensively
(Purlis, 2010; Dong and Hu, 2011). However, the results of
conventional extraction approaches were unsatisfactory in this
case because the size of a chicken was changing slightly in the
roasting process and the background is also disturbed by vapor
and splashing residues in the roasting process due to changes of
temperature and humidity. Therefore, in our study, the Gaussian
Average method based on the background subtractions
(Massimo, 2005) is used as an auxiliary tool to obtain a rectangular
mask and this mask is used as an initial condition at first frame for
creating a sub-image for the proposed segmentation approach.
This mask is marked as a sub-image in all subsequent frames,
tightly wrapped against the food in video processing in order to
reduce the computing cost in segmentation.
And then, a new level set based approach is proposed
for extracting the chicken area in this paper. In recent years,