Improved infrared thermal imaging cervical vertebra part extraction method based on Yolo v3

Abstract

The invention discloses a Yolo v3-based improved infrared thermography cervical vertebra part extraction method, which comprises the following steps of: acquiring an infrared thermogram of a cervical vertebra part, then in an upper computer, scaling the length and width of the infrared thermogram to be a multiple of 32, then carrying out noise reduction processing on the scaled infrared thermogram through a Gaussian filtering algorithm, and finally, extracting the cervical vertebra part from the infrared thermogram. And inputting the noise-reduced picture into an improved Yolo v3 target detection network for target extraction to obtain an image with a prediction bounding box, category and prediction confidence of the cervical vertebra part, and displaying the image in an upper computer. According to the method, the residual feature enhancement module is introduced into the Yolo v3 target detection network to reduce high-level information loss caused by feature fusion, on the other hand, the attention mechanism module is introduced, rapid scanning is performed at a low level, a target area needing to be focused is obtained, the semantic difference between the target area and high-level features is reduced, the recognition precision is improved, and the recognition efficiency is improved. And meanwhile, the advantages of high generalization ability, high speed and the like are reserved.

Description

technical field [0001] The invention relates to the related fields of computer vision and target detection, in particular to a method for extracting cervical vertebrae parts from infrared thermal images improved based on Yolo v3. Background technique [0002] At present, cervical spondylosis has become a frequently-occurring disease. Clinicians use Nuclear Magnetic Resonance Imaging (NMRI) and Computed Tomography (CT) perfusion images to evaluate the condition and severity of cervical spondylosis. This method Not only is it time-consuming, it is easy to delay the treatment, and it requires a very experienced doctor, otherwise it is easy to cause inaccurate diagnosis of the disease. [0003] With the development of computer technology and the wide application of computer vision principles, the use of computer image processing technology to detect or segment targets in real time is becoming more and more popular. In the fields of intelligent transportation systems, intelligent...

AI Technical Summary

Problems solved by technology

Although the one-stage target detection algorithm Yolo v3 has obvious advantages in real-time detection, there are two problems in Yolo v3 using feature pyramids for feature fusion, which affect the target detection accuracy of Yolo v3:

[0004] Loss of high-level features: High-level features are adapted to low-level features through convolution and upsampling. This process will lose certain information because the number of channels decreases during the adaptation process;

Images