An Algorithm for Extracting Image Features Using a Random Deep Neural Network

In this paper, a random deep convolutional neural network-based technique for extracting image features is devised and put into practice. The image features are also richer following the image preprocessing procedure described in this article. Additionally, more consistent feature points are recognized, duplicate local features are deleted, and this prevents the feature matching from encountering too many unmatched pairs. Since the self-encoder can successfully rebuild the training data set, data that is close to the training set, and the data itself, this work offers some original ideas for constructing the model. The study's findings have demonstrated the successful outcomes.