Image Feature Extraction under Microscope Based on Visual Saliency

  • Guoshan Liu
Keywords: Microscope, Image Feature Extraction, Visual Saliency.


As there are a great variety of microorganisms, the backgrounds of microorganism images acquired vary
greatly, making microorganism image detection and feature extraction a complex and arduous task. It is
estimated that the reliability of visual saliency may be able to properly accomplish, without priori
knowledge, some image processing, including image segmentation, object recognition, adaptive
compression and the like. This paper studies the object extraction algorithm of complex microorganism
image based on visual saliency. For a saliency map, this algorithm is to highlight salient regions and inhibit
non-salient regions as far as possible to make the brightness contrast between these two kinds of regions all
the more apparent. Besides, this paper also optimizes the original features extracted and presents a
microorganism image detection algorithm based on visual saliency of regional contrast, which can make an
assessment of global contrast and spatial consistency of microorganism image and adaptively make a
judgment whether to emphasize local detection or global detection or somewhere in between. The test
experiment proves that the algorithm of this paper works very well in microorganism image feature