Image Feature Extraction Algorithm based on Random Deep Neural Network

摘要

Image feature extraction algorithm based on the random deep neural network is designed and implemented in this paper. After the image preprocessing method in this article, the image details are also richer, more uniform feature points are detected, and redundant feature points are eliminated, avoiding too many mismatched pairs in the feature matching. The self-encoder has good reconstruction ability to the training data set and the data similar to training data set, and the reconstruction effect to the data, hence, this paper provides the novel ideas to then construct the model. The experiment results have proven the satisfactory results.