<![CDATA[Based on the existing image mosaic technology, this paper introduces the granular computing and obtains a simplified new algorithm. The image mosaic executed by this algorithm at first establishes correlation model on the basis of granular computing theory, and obtains edge map of each image needing mosaic. The new calculation method is used to calculate gradient of in different columns of edge map, to obtain the feature point coordinates with the maximum gradient; meanwhile, all feature points of two images are matched with each other, to acquire the best matching point. In addition, the error-correcting mechanism is introduced in the matching process, which is used to delete feature points with matching error. The correlation calculation is carried out for the matching pixels acquired by the above processing, to get the feature transformational matrix of the two images. According to the matrix, two separated image maps map into the same plane. The slow transitional mosaic method is applied in the aspect of image addition plus overlap removal, so that images have no bulgy boundary after mosaics. The whole image mosaic process shows that the given granular computing algorithm is superior to the traditional one both in the number of processed images and the number of processing, and the mosaic image gained has high quality.]]>
Copyrights © 2013
