Feature Extraction and Target Optimization Model Construction of Micrograph
After years of development, microimage has penetrated into various fields, playing an important role in
scientific research, medical and health care, industrial production, film and television art, etc. The purpose of
this paper is to study the feature extraction and target optimization model construction of microscopic image.
Firstly, the method of micro image acquisition is described, the advantage of genetic algorithm is understood,
and the feature extraction based on genetic algorithm and the model construction based on target optimization
are studied. In this paper, MSE, PSNR and SSIM are used to quantitatively analyze and compare 100 randomly
selected images in the ImageNet verification set. The experimental results show that the PSNR of the
quantitative model (target optimization) in the case of target extraction failure is the second 18.68, and the SSIM
is the first 0.66. In the case of target acquisition failure, the performance of the algorithm will appear slight
degradation, slightly better than the generative model in PSNR measurement standard, but still better than the
global and local discriminators and partial convolution model. The quantitative analysis results of the improved
model in image restoration are better than other models.