An image generation method and system based on edge closure and commonality detection

Abstract

The invention discloses an image generation method and system based on edge closure and commonality detection. The method includes: acquiring input text, and screening the text for text elements; respectively inputting the screened text elements into an image search engine, and outputting corresponding Image search results; multiple effective images are screened out, and edge detection is performed on each image to obtain an edge detection image; the edge closure operation is performed on the entities in the edge detection image by the nearest neighbor detection connection method, and an entity with a closed edge is obtained; All the entities in the image are cropped according to the corresponding closed edges to obtain the candidate pure entity images; the candidate pure entity images are classified through the unsupervised image classification algorithm, the total number of images in each category is counted, and the category with the largest total number of images is classified as The image in is used as the semantic image corresponding to the text element. The present invention can generate high-quality, pure entity semantic images that conform to common sense and have a transparent or pure color background based on the text.

Description

technical field [0001] The invention belongs to the technical field of text-image generation, and in particular relates to an image semantic segmentation method, system, device and storage medium based on pixel closure and commonality detection. Background technique [0002] In the field of content production, the efficient production of IP determines the competitiveness of enterprises with IP production as their core business. However, because IP production is a typical creative work, the deep learning technology developed rapidly in the past ten years cannot be well adapted to the task of "creation". The application of deep learning in images is mainly in the field of machine vision, and the task of converting text to still images is a typical type of artificial intelligence technology that increases the amount of information and requires neural networks to have creative capabilities. The underlying principle is to use convolutional neural network or cyclic neural network...

AI Technical Summary

Problems solved by technology

[0006] In view of this, the present invention proposes an image generation method, system, device, and storage medium based on pixel closure and commonality detection, which are used to solve the cumbersome manual calibration and training in the existing text-image generation technology and the inability to generate images based on text. The problem of plausible semantic images

Images