Image processing method, electronic equipment and storage medium

Abstract

The invention provides an image processing method, electronic equipment and a storage medium. The method comprises the steps of carrying out the analysis of to-be-processed images of different preset sizes according to a pre-trained first image processing model, and obtaining a ternary image set of the to-be-processed images, inputting the ternary image set and the target training data set into a pre-trained second image processing model for training to obtain a third image processing model, analyzing the to-be-processed image based on the third image processing model to obtain a foreground mask of the to-be-processed image, and performing target background filling on the foreground mask to obtain a target image. The acquisition efficiency of the ternary image and the use flexibility of the image processing model can be improved.

Description

technical field [0001] The present application relates to the technical field of image processing, and in particular, to an image processing method, an electronic device and a storage medium. Background technique [0002] At present, the big data model is usually used for image processing, such as image matting processing, etc. However, the training process of the big data model requires a large number of original images and ternary maps of the original images. However, the existing methods for obtaining the original image ternary map mainly use manual labeling. Due to the low efficiency of manual labeling and the inability to guarantee the accuracy, it is difficult to obtain training samples for big data models. And because the corresponding ternary images of different images cannot be obtained uniformly, different big data models need to be trained for different application scenarios. Therefore, there are problems in that it is difficult to obtain training samples and can...

AI Technical Summary

Problems solved by technology

[0002] At present, big data models are usually used for image processing, such as image matting processing, etc., but because the training process of big data models requires a large number of original images and triplet images of original images

However, the existing method of obtaining the original image ternary map is mainly marked manually. Due to the low efficiency of manual labeling and the inability to guarantee accuracy, it is difficult to obtain training samples for big data models.

And because the corresponding ternary images of different images cannot be obtained uniformly, different big data models need to be trained for different application scenarios

Therefore, when using big data models for image processing, it is difficult to obtain training samples and cannot be flexibly applied.

Images