Noise reduction method for one-dimensional signal of single-pixel imaging system in low-light environment

Abstract

The invention discloses a noise reduction method for a one-dimensional signal of a single-pixel imaging system in a low-light environment. A noise reduction convolutional neural network is trained by using simulation data, and noise reduction processing is directly carried out on a one-dimensional bucket detector signal: firstly, the property of noise in a low-light environment is researched, a real noise model is generated, a series of simulated noise-containing signals are generated by using the noise model, the simulated noise-containing signals and a simulated noise-free signal are used for training the noise reduction convolutional neural network, and real experimental data are used for testing. According to the invention, the network is trained by analog data, so that the data acquisition cost is reduced; the noise-containing signals obtained by short integral sampling can be quickly reduced, a reconstructed image with better quality can be obtained, and meanwhile, the imaging efficiency is improved; and the development of deep learning in the field of single-pixel imaging in the low-light environment is promoted.

Description

Technical field [0001] The present invention belongs to the field of image processing, in particular a noise reduction method of one-dimensional signal of single pixel imaging system in a micro-light environment [0002] technical background [0003] Single-Pixel Imaging is a method using associated measurement reconstructed images, which is rapidly developed as a emerging imaging technology. Single pixel imaging technology does not need to be probed using any array detectors, which brings a potential ability to solve some of the challenges that are unable to deal with traditional imaging. For example, single pixel cameras allow people to build a low-cost imaging system that can work at an array detector. In the past decade, single pixel imaging techniques have a lot of demo in multi-spectral imaging, radar imaging, 3D imaging and real-time imaging. [0004] In recent years, researchers have solved some problems in single pixel imaging by combining deep learning with single pixel ...

AI Technical Summary

Problems solved by technology

However, how to simulate the behavior of noise, analyze the nature of noise and generate simulated noise-containing signals in low-light environment has not been studied in depth.

Images