A method and apparatus for image cropping

Abstract

The invention discloses a method and a device for image clipping, which relates to the technical field of computers. One embodiment of the method includes: generating a clipping frame according to a size of an image to be clipped and a target size; Using a clipping box to select an image block to be clipped; evulating The selected image blocks by using the composition quality evaluation model, andtaking the image blocks with the highest evaluation score as the target images. This method can solve the problem of multi-agent image clipping, fully consider the rationality and aesthetics of imagecomposition, so as to obtain the most reasonable region of global composition.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to an image cutting method and device. Background technique [0002] With the development of multimedia and mobile terminal technologies, a large number of rich pictures are generated every day. Whether it is ordinary users, magazine editors, new media editors or enterprise operators, they all need to crop pictures according to specific needs. On the one hand, it makes the image composition more reasonable. On the other hand, it needs to meet certain display size requirements. Image display is performed by cropping and transforming the image. [0003] Most of the currently commonly used image cropping methods are automatic cropping based on subject recognition (please refer to the patent application whose application number is CN107610131A). and the center position of the subject image to determine the cropping position to crop the image. [0004] However, the automatic...

AI Technical Summary

Problems solved by technology

[0005] 1. In this solution, only target sizes that meet certain cropping conditions can be cropped, and arbitrary target sizes cannot be processed;

[0006] 2. The idea of ​​determining the cropping position by the subject of the image (person, animal, flower) is controversial: (1) how to define the subject, and how to deal with the category of subjects that have not appeared before; (2) for multi-subject images, the solution needs to be determined The core subject has certain problems in technical implementation, and cannot guarantee the integrity of information and the rationality of the overall composition; (3) The reasonable cropped area of ​​​​the picture is not more reasonable to display with the subject, but should be related to the content and composition of each image ;

[0007] 3. The composition rules of this scheme are all determined by the golden position points of the fixed Jiugongge composition, which has nothing to do with the image content and composition. Therefore, the main image determined during cropping is not accurate enough, resulting in incomplete cropping information

[0008] In summary, the cost of image cropping by manpower is relatively high, and the image results obtained by the automatic cropping technology based on subject recognition at this stage are not satisfactory, and there are situations where the cropping information is incomplete and the cropping area is unreasonable

Images