A preprocessing is a process for preparing an input image to be suitable for processing (i.e, segmentation
and feature extraction). The first preprocessing process is to resize an input image. Since sizes of our
input images are largely varied, we need to resize them into the same scale in order to correctly compute
algae shape features, particularly, shape measurement features (will be described in Section 3.3). The
images whose longest side is larger than 400 pixels are resized to 400 pixels while the other side remains
in the same aspect ratio.
The second preprocessing process is to convert a color input image into a gray-scale image. We
perform color-to-gray image transformation because we do not use color information of algae in a
classification process. The first reason is that our algae images are produced from several imaging
systems. Thus, colors of algae of the same genus may be varied significantly, depending on imaging
systems and illumination adjustment. Secondly, colors or pigments of algae depend on environmental
conditions in which they are growing. Colors of algae of the same genus may vary in a wide range,
while colors of algae of different genera may be identical. As a result, color feature of algae is not
suitable for identifying or discriminating algae.