A spot segmentation method of low contrast and strong noise based on NSCT and 2D histogram

Spot segmentation is an important section of detecting coal level of coal bin in computer vision. When it is difficult to proceed, analyze and locate the spot position for the image obtained under the condition of low contrast and strong noise. This paper aim at achieving spot segmentation in complex underground, a novel spot segmentation method based on nonsubsampled contourlet transform (NSCT) and 2D histogram is proposed. First, the original image is decomposed into different scales and directional subbands by using NSCT. Then, the coefficients are adjusted by adopting differently adaptive transform functions to different subbands and by adopting differently adaptive threshold and spot-sensitive factor. Last, the extraordinary segmentation effects are achieved by 2D histogram threshold segmentation method through the spatial correlation of gray level and the flexibility as well as efficiency of the threshold searching of multi-resolution threshold segmentation method. Simulating results indicate that this method can efficiently implement spot segmentation, better keep spot structure and preserve spot textures and details.