Implementation of a combined approach using autoencoder and Kalman filter for image processing: a numerical experiment

In the paper, the problem of predicting the state of moving objects under the conditions of variable environment and exposure to noise is investigated. A combined method based on the Kalman filter and autoencoders is proposed to improve the accuracy and stability of the predictions. The Kalman filter provides object state estimation and data filtering, while autoencoders extract key features from object images. The proposed method improves the accuracy and robustness of the predictions, especially under variable environment and exposure to noise.