2.5. Image processing
Background segmentation was first performed on the
original images. Photometric differences between background
and rib-eye were used to develop discriminant functions for
background removal. In this experiment, a boundary tracking
algorithm was used to conduct this operation. Details of this
algorithm are given by Wang[13].
After removing the background, the rib eye image contained
l.d. muscle, subcutaneous fat surrounding muscle and
intramuscular fat referring to the fat flecks (marbling)
surrounded by l.d. muscle. Because the lean color scores are
defined by l.d. muscle color, separating the l.d. muscle from
the extraneous tissues was necessary. To achieve this
objective, morphological and logical operations were
investigated. The procedures are summarized as follows:
(1) For each rib-eye image, the optimum threshold value was
computed;
(2) Rib-eye images were binarized according to the obtained
optimum threshold value;
(3) The component labeling algorithm, erosion and dilation
operations were used to remove the extraneous tissues while
the binarized l.d. image was identified and remained;
(4) A logical operation ‘AND’ of the binary l.d. image with
the original image resulted in a colored l.d. image;
(5) The shareholding algorithm was used again and the
intramuscular fat flecks within l.d. were removed;
(6) A resultant l.d. muscle image was obtained for estimating
and analysis of color features.
Each sample image was preprocessed in the same way and
all of resulting images were subjected to extraction and
analysis of color features.
sample surface as showed in Fig. 1.