Review of image recognition systems for noisy environments
Junkai Lu · Applied and Computational Engineering · 2023
One of the major challenges faced by image recognition systems in practical applications is the presence of noise in the physical world. To address this challenge, this paper proposes two different approaches. The first approach involves constructing a noisy dataset and training the image recognition system to tolerate noise. The second approach utilizes a combination of denoising methods and pre-trained image recognition systems. Experimental tests and analyses are conducted to evaluate the advantages and disadvantages of each approach. Additionally, this paper investigates the relationship between different noise reduction methods and their impact on improving the recognition rate and resource consumption of images containing Gaussian noise, specifically when neural networks are used for recognition.