Research on the Recognition and Retrieval Model of Ancient Coin Images Based on Deep Learning

In the field of ancient coin image research, researchers usually use local features to describe ancient coin images, which requires manpower and time to extract features from ancient coin images. In practical applications, complex processing is performed on a retrieved ancient coin image before retrieval, which is obviously inefficient. In response to such issues, this paper utilizes convolutional neural network to extract deep features of ancient coin images, improving the model’s semantic feature representation ability of ancient coin images. While learning image semantic features, map them into compact binary hash codes to improve retrieval efficiency. This paper designs an ancient coin image retrieval model based on convolutional neural network for the ancient coin dataset. A convolutional neural network was used to construct the model which can effectively extract semantic features of ancient coin images and store representations with fewer hash codes. Through experimental simulation, it was verified that the ancient coin image retrieval model based on convolutional neural network in this paper can achieve better performance than traditional feature extraction methods in various evaluation indicators.