Image generation method and device

Abstract

The invention discloses an image generation method and device. The method comprises the following steps: acquiring a to-be-processed video and a target clothing image; extracting at least two continuous video frame images from the video to be processed, wherein the video frame image comprises a target object; performing target object segmentation processing on the at least two video frame images to obtain a binary image of the target object; and fusing the target clothing image, the binary image and a target video frame image in the at least two video frame images to obtain a target fused image. The target object can still be accurately matched with the specified costume picture under the condition that the background is complex, and a more natural identification photo is output.

Description

technical field [0001] The invention relates to the technical field of computer image processing, in particular to an image generation method and device. Background technique [0002] At present, people not only need to spend money to buy or rent formal clothes, but also need to go to professional ID photo shooting places to take photos, which is time-consuming and expensive. With the rapid development of deep learning technology, the ID photo generation method based on deep learning can automatically generate ID photos, saving time and effort. [0003] In the prior art, the traditional image segmentation method and matting are used to synthesize ID photos on a single frame image. This method distinguishes the front and back backgrounds through fixed mathematical operations, which is easily affected by the complexity of the background. In some non-solid color backgrounds, The foreground and background are prone to crossing and misclassification, which greatly affects the fi...

AI Technical Summary

Problems solved by technology

[0003] In the prior art, the traditional image segmentation method and matting are used to synthesize ID photos on a single frame image. This method distinguishes the front and back backgrounds through fixed mathematical operations, which is easily affected by the complexity of the background. In some non-solid color backgrounds, The foreground and background are prone to crossing and misclassification, which greatly affects the final composition effect, and the final composition effect through cutout is only background replacement, not clothing replacement. If the ID photo requires formal attire, you still need to wear formal attire

[0004] In addition, when the background is relatively complex, such as the color of the hair around the hair is similar, the image is exposed, blurred, etc., the segmented human body area is mixed with the background area, and the effect is not good. At this time, the segmented human body area is fused with the clothing image Poor effect and other issues

Images