Panoramic image real-time stitching method and panoramic image real-time stitching system based on mobile terminal

Abstract

The invention discloses a panoramic image real-time stitching method and a panoramic image real-time stitching system based on a mobile terminal. The method comprises the following steps: detecting the feature points of a current frame of an acquired video stream; carrying out feature point matching on an obtained feature point attribute sequence of the current frame and a feature point attribute sequence of a previous frame of the video stream; calculating a transformation matrix between the current frame and the previous frame according to the result of matching; and finally, stitching the current frame and the previous frame in real time according to the transformation matrix, repeating the cycle, sequentially getting a preliminary panoramic image, an intermediate panoramic image and a final panoramic image, and previewing and displaying the preliminary panoramic image or the intermediate panoramic image or the final panoramic image in real time. The algorithm is simple, and there is a small amount of calculation. A panoramic image can be stitched and previewed in real time in the process of panoramic shooting. The method and the system have low requirement on hardware configuration, and are particularly applicable to high-resolution real-time panoramic shooting in mobile terminals with medium or low configuration such as mobile phones.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a mobile terminal-based real-time mosaic method for panoramic images and a system for applying the method. Background technique [0002] The current panoramic stitching technology on mobile devices generally generates panoramic images by collecting images from multiple angles and performing computationally intensive operations such as image registration and fusion on the original high-resolution images according to the feature point matching method. [0003] First of all, due to the large amount of calculation for image registration and fusion, images can only be collected by interval sampling. Due to the large interval of image acquisition, first of all, the overlapping part is small, which is not good for the alignment of feature points, and it is easy to produce ghosting of the panorama. [0004] Secondly, due to the large difference between the sampled scenes, usual...

AI Technical Summary

Problems solved by technology

Due to the large interval of image acquisition, firstly, the overlapping part is small, which is not good for the alignment of feature points, and it is easy to produce ghosting of the panorama

[0004] Secondly, due to the large difference between the sampled scenes, usually the white balance and exposure parameters of the camera have undergone relatively large changes, so the post-processing part generally needs to use some exposure compensation and complex methods to mix images to make the image transition natural. Make users wait longer for panoramic images

[0005] Again, this collection method determines that it is difficult for mobile devices to generate some intermediate results for users to preview in real time, making users unpredictable the final stitching results

[0006] In addition, even if the preview of the stitching result is provided, due to the large amount of calculation, it is impossible to obtain a smooth preview effect, and it cannot meet the real-time performance of the preview result. The final fusion result usually has some unnatural jaggedness and smearing of the fusion area, and , higher requirements for hardware configuration

Images