Semantic-based image adaptive method by combination of slit cropping and non-homogeneous mapping

Abstract

The invention provides a semantic-based image adaptive method by combination of slit cropping and non-homogeneous mapping, comprising the following steps: by combining with a semantic analysis result of an intermediate layer of a sports video, extracting statistical characteristics for subjective evaluation on semantic region importance of a user for an image frame, and weighing an underlying feature-based energy function through the statistical characteristics to obtain the energy function after semantic weighing; removing unimportant information in images by using a slit cropping method for positive energy, and meanwhile measuring important information deformation according to variation of an important semantic edge; and when deformation exceeds a set index, terminating the slit cropping method, and using a non-homogeneous mapping method to obtain the images of a target size, wherein, a method framework is as shown in an attached figure of an abstract. The semantic-based image adaptive method organically combines the existing two relatively good methods-the slit cropping method and the non-homogeneous mapping method, gives full play to respective advantages to achieve a comprehensive optimum image video adaptive result, and introduces energy function calculation after semantic weighing to achieve adaptive image sizes based on the semantic content.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to the research and realization of an image self-adaptive method based on the combination of semantic-based thin seam cutting and non-uniform mapping. Background technique [0002] With the popularity of mobile devices, terminal devices suitable for different networks such as TVs, notebook computers, PDAs, and mobile phones are constantly being updated, and display terminals of different types and sizes are emerging to meet various user needs. In order to ensure that users with different devices can watch the same image content comfortably, it is required that the image content can be adapted to user terminals of different sizes and proportions. It is of great application significance to study the adaptive technology that keeps the main content of the image. [0003] At present, the main image size adaptive methods include linear deformation (Scaling) method, cropping (Crop...

AI Technical Summary

Problems solved by technology

The bottleneck of this type of method is that the universal object segmentation technology and the robust foreground and background classification technology are not yet mature, which also leads to limitations in the application of this type of method

Images