Dynamic tracking method for multi-frame moving target based on symmetry algorithm

• Published in: Journal of intelligent & fuzzy systems 2020-01, Vol.38 (6), p.7693-7703
• Main Author: Ou, Yufeng
• Format: Article
• Language: English
• Subjects: Accuracy; Adaptive algorithms; Algorithms; Encryption; Image classification; Moving object recognition; Moving targets; Pixels; Symmetry; Target recognition; Tracking
• ISSN: 1064-1246; 1875-8967
• DOI: 10.3233/JIFS-179839

Aiming at the problems of poor tracking effect and low precision of traditional methods, a symmetric algorithm is introduced to dynamically track moving objects in multiple frames. Symmetric encryption algorithm is used to hide the reversible information of the image, and the correlation between adjacent pixels of the encrypted image is increased through high-bit flip prediction. Generate sparse pixel classification matrix to improve compression efficiency and reduce the occupation of effective bit-plane space. Embed secret information in the most significant bit of the encrypted pixel based on the hidden key to generate an encrypted-portable image. The receiving end can generate two kinds of decrypted images, one is consistent with the original image, and the other contains the secret information. Has hidden key to extract secret information from ciphertext image and plaintext image separately. Under the hiding of target image information, a window adaptive adjustment algorithm is introduced to track the target, and a comprehensive predictor is designed to predict and track the target when the target is temporarily blocked. The experimental results show that the target tracking accuracy of this method can reach 99% and the prediction accuracy can reach 100%. The proposed method has strong security performance and high tracking prediction accuracy, and is a feasible target dynamic tracking method.