Underground Image Mosaic Technique Based On Image Enhancement

Many images will be affected because of the complex Coal mine environment. Similar background color and the target color and most light source lighting partly.make captured images divide between the brightest zone and darkest zone. When extracting the feature point, due to the low noise ratio in the dark region, effective feature points are nearly undetectable, making trouble in image registration step. An improved Retinex algorithm based on bilateral filtering was proposed. It can enhance the images in image pre-processing stage. Thanks to the Bilateral filtering algorithm, the edges are well fully protected, and the filtered image is still very smooth. The light area and dark area can be separated on this basis. Enhancing and filtering the dark area alone can avoid dealing with the bright and dark areas simultaneously, which can meet the requirement. After enhancement, calculate the feature point descriptor vectors by SURF algorithm, use the closest distance than the next closest method to match feature points matching. Finally, use feature points and a linear gradient fusion method to fuse the image on the calculated transformation matrix. Comparing the number of feature points before and after enhancing the figure, enhancement algorithm was verified effective. Besides, the algorithm significantly improves splicing efficiency of the SURF (Speeded Up Robust Features) algorithm, and improve the accuracy and rapidity matching stitching.Simulation results show that this algorithm for image pre-processing the image not only meet the needs of the human eye, but also significantly increase the number of valid feature points in the dark areas of the image. Using the method proposed in this essay to enhance the figure with obvious separated edge between the brightest and darkest area firstly before using SURF algorithm to extract feature points, the figure containing the same area can be spliced into a clear, wide view images quickly and accurately.