Hybrid of Mean-shift and median-cut algorithm for fish segmentation

In this paper, a hybrid of mean-shift and median-cut algorithm was introduced to perform segmentation on color images. Firstly, the image pre-processing technique was applied to enhance the image before changing the color space to LUV color space. Then mean-shift segmentation was introduced. However, some region has no semantic meaning since the mean-shift algorithm performed low level segmentation. To overcome this problem, median-cut algorithm was proposed to join the gap left by the mean-shift. The hybrid of the two methods has improved the contour extraction for further processing.