A small sample terahertz image foreign matter detection method based on integrated deep learning

Abstract

The invention discloses a small sample terahertz image foreign matter detection method based on integrated deep learning, and mainly solves the problems that an existing method needs to manually design image features, the training process is complex, and foreign matter detection cannot be carried out on a small sample terahertz image with a small sample number. The method comprises the following specific steps: (1) making a small sample terahertz image data set; (2) amplifying the image training set; (3) constructing an integrated deep learning network; (4) training the integrated deep learning network; and (5) detecting the image test set. According to the method, the image features can be automatically extracted, the training process is simple, the situation that a certain type of samples in actual samples are particularly few is considered, foreign matter detection can be conducted on the terahertz images of the small samples, and the detection accuracy of the certain type of samples in the small samples is particularly few can be improved.

Description

Technical field [0001] The present invention belongs to the technical field of image processing, and further relates to a small sample terahertz image foreign body detection method based on integrated deep learning in the technical field of target detection image processing. The invention can be used to detect foreign objects such as knives, explosives, and mobile phones hidden by the human body from the terahertz image. Background technique [0002] In recent years, with the frequent occurrence of terrorist attacks at home and abroad, rapid and efficient human foreign body detection in crowded public places such as airports, railway stations, and subways has become an urgent need under the current severe security situation. Because terahertz wave is between infrared and microwave, it has some special properties that X-ray, light wave / infrared and microwave do not possess, making terahertz imaging very suitable for human body foreign body detection. Compared with traditional met...

AI Technical Summary

Problems solved by technology

Although this method can reduce the false alarm rate and improve the accuracy of target detection by constructing a spatial distribution histogram and a human scale model, and the structure of the millimeter wave image is similar to that of the terahertz image, this method can be used for foreign object detection of the terahertz image, but , the disadvantage of this method is that it needs to extract the contour map of the human body, construct a histogram vertical to the horizontal space distribution, preset the proportion of the human body, etc., which makes the foreign object detection process very cumbersome, and does not consider the distribution of small samples In fact, it does not increase the number of samples, and it does not consider the small sample situation of a certain type of sample in the actual sample, so this method cannot detect foreign objects in small-sample terahertz images.

However, the disadvantage of this method is that it needs to establish a statistical model based on generalized compound distribution and generalized K distribution. Sample situation, this method is not suitable for foreign object detection on small sample terahertz images

Images