X-ray sample image generation method, X-ray sample image generation equipment and storage device

Abstract

The invention discloses an X-ray sample image generation method, X-ray sample image generation equipment and a storage device. The X-ray sample image generation method comprises the steps of obtaininga to-be-fused background image and a to-be-fused foreground image comprising a first to-be-identified target; obtaining a first sub-foreground image containing a first to-be-identified target from the to-be-fused foreground image, and extracting a first sub-background image matched with the size of the first sub-foreground image from the to-be-fused background image; inverting the pixel values ofthe first sub-background image and the first sub-foreground image to obtain a second sub-background image and a second sub-foreground image; performing weighting processing on the second sub-background image and the second sub-foreground image to obtain a first fusion image; inverting the pixel value of the first fusion image to obtain a second fusion image; and generating a first X-ray sample image for training the target recognition model by using the second fusion image and the to-be-fused background image. Through the above method, an effective sample image is provided for training the target recognition model.

Description

technical field [0001] The present application relates to the technical field of image processing, and in particular to a method for generating an X-ray sample image, a generating device and a storage device. Background technique [0002] The imaging principle of X-ray images is: X-rays have different penetrability to different substances and are attenuated to different degrees. The detector receives the attenuated X-ray signals and converts them into images for display after signal processing. [0003] X-ray images are very different from conventional images: conventional images have surface texture information, but X-ray images have no texture information, only object physical density information; conventional images have occlusion effects when superimposed, but X-ray images have perspective. Therefore, the technical difficulty of X-ray image target detection is not small: a good X-ray image target detection algorithm is inseparable from a well-conceived deep learning algo...

AI Technical Summary

Problems solved by technology

Although the X-ray security inspection machines used in the security field generate a large number of X-ray images every day, the proportion of positive sample images containing contraband to be identified is very low and it is difficult to screen. Therefore, a large number of effective X-ray image training positive sample images are constructed. There is a certain degree of difficulty in the training set

Images