An Automatic Collage Method of Hierarchical Atlas Based on Image Importance

Abstract

An automatic collage method for hierarchical atlases based on image importance consists of four steps. Step 1: Use the Meanshift algorithm to segment the image, extract the salient areas in the image in a rectangular manner, and merge them into one if there are multiple salient areas; Step 2: Through the color complexity and pixel gradient complexity of the foreground and foreground of the image Calculate the degree of difference between the front and back of the image, and then calculate the mean value of the difference between the image and other images in the atlas to obtain the representativeness of the image. Finally, the average of the difference between the foreground and the background and the difference between the images can get the respective importance of each image. Degree; Step 3: Save the hierarchical relationship of the atlas with an XML file, and use the image importance obtained above as the weight of the node and save the circumscribed rectangular outline of the salient area; Step 4: Use the Voronoi tree diagram to visualize the atlas hierarchy, canvas area Indicates the importance weight of the atlas subset or image, and automatically collages the salient areas of each image according to the rectangular outline of the salient area of ​​the node and the image scaling technology.

Description

technical field [0001] The invention relates to an atlas processing technology, in particular to an automatic image collage method with excellent visual effect. Background technique [0002] With the improvement of the performance of smartphones and the upgrade of camera pixels, people are more and more used to recording the details of life by taking pictures casually. How to browse and manage more and more photos has become a problem that people pay attention to. Although various pictures, photo-sharing software and websites emerge in an endless stream, enabling people to share photos with others. However, a large number of images make it difficult to browse and retrieve, and the commonly used classification storage method also makes it possible to only see the folder name when viewing a certain type of image, but cannot obtain specific image information. The image collage technology aims to collage multiple images in one canvas, allowing users to get more image informati...

AI Technical Summary

Problems solved by technology

Some automatic image collage technology solutions have been proposed, but there are certain shortcomings, such as the number of supported images is small, generally 20 to 40, and too many images will lead to poor visual effects. And cannot support hierarchical atlas data

[0003] For example, Microsoft provides an image collage software AutoCollage, which can extract the ROI area in the image, but it is time-consuming, and the shape division of the canvas is relatively simple

In the literature, Yu Zongqiao proposed to use the circle layout method for image collage. It calculates the importance of the image separately, and then divides the canvas by circumscribed circles. The disadvantage is that the calculation method of image importance is not effective and the canvas utilization rate is low.

In the literature, Liu Tie et al. proposed a collage method that distinguishes different important areas of the image, and then occludes non-important areas from each other. This method allows the canvas to mainly present the ROI area in the image. The disadvantage is that the view is more chaotic, and the lower layer There is a high probability that the picture is blocked

Brian et al. proposed a layout scheme called BRIC (blocked recursive image composition), which takes less time but has a single collage effect

[0004] The image collage scheme mentioned above does not achieve a good balance between the collage effect and efficiency, and only considers a small number of image collages on the same layer, and does not support large-scale multi-level multi-theme atlases. Collage Visualization Scheme

Images