A High Fidelity Mosaic Technology For Large Mural Image Based On SFM

Relic is a valuable historical and cultural heritage of great historical and artistic value. Because of factors such as aging and destruction by human, relics protection work is facing many difficulties. As a class of common heritage, Ancient murals can be preserved through digital approach in addition to the conventional means of protection. The digitization of ancient mural requires high accuracy and high resolution. The common practice is to photograph partly, and then make the panorama formation by means of image mosaic. Research group has solved the problem of image block acquisition, work is now focused on the image mosaic.The existing image mosaic method is not satisfying the demand of murals mosaic in the mural mosaic work, especially in the high image resolution and large number of pictures, the overall image tends to be different with the reality, and splicing speed is also not satisfactory. This paper worked to find the right ways to improve the speed of image mosaic murals and the accuracy of the final result, to meet the needs of high fidelity.According to the principle of murals capture, the paper presents a reliable feature point matching method based on physical location information which can effectively reduce the target area in the matching process for feature points, thereby enhancing the speed and accuracy of matching and lay a good foundation for the follow-up image mosaic work.This paper presents an image mosaic technology based on SFM, which uses SFM algorithms to estimate the three-dimensional coordinates of feature points in the image, and then use spatial information to calculate a suitable projection plane and do re-projection for all images to be spliced. This technology can effectively reduce the error of final overall big picture caused by the uneven wall paintings and the distortion of images, and optimize mosaic results.The two methods proposed in this paper, when used in the image mosaic system, can facilitate large-format murals mosaic. The final experimental results show that the proposed method can effectively reduce the errosr caused by objective reasons more than any other methods, and therefore meet the requirements of fidelity, and achieved the desired results.