The neural network is well-trained and tested by simulated space particle holographic reconstruction samples which are prepared as a dataset before training. The method is based on the theory of deep learning and we build a deep learning neural network. In view of the above situation, we propose a denoising method for space particle holographic detection. If the latter method uses inappropriate filtering method, it will add extra errors that result in the loss of reconstruction results information. Although the former can play a good denoising effect, it takes a long time to iterate many times in a procedure. Actually, the traditional methods have some disadvantages. The other is to directly filter and denoise the reconstruction results containing noise by filtering algorithm. One is to improve the quality of the reconstruction results by controlling the threshold of the iterative shrinkage reconstruction algorithm, so as to achieve the purpose of noise reduction. In order to obtain the multi-layer reconstruction results without noise, there are two kinds of traditional denoising methods. It can realize reconstruct multi-layer of the measured object, that is, it can reconstruct multi-layer cross sections of the measured object.įor each layer of cross sections, because of the influence of holographic recording and reconstruction process, there will be spatial particle distribution information and noise on it. This technology can be used to record and reconstruct space particles. For example, holographic technology is commonly used in optical NDE. So as to avoid affecting the state of the measured objects, optical nondestructive examination (NDE) is a very efficient method. In the industrial field, it is necessary to detect the concentration and the size of spatial particles field. Spatial particles refer to suspended particles in gases or liquids, such as PM2.5 and droplets in the air.
0 kommentar(er)
