Improvement of Attribute Reduction Classification Algorithm Based on Rough Sets

Since rough sets cannot deal with noisy image information, there will be inaccurate classification problems in image classification, and other algorithms must be combined to achieve accurate classification results. Although the neural network classification method can better extract image features, when the spatial dimension of the image is too large and the training time increases, it will also lead to low image classification accuracy. Therefore, this paper combines rough set(RS) theory and neural network(NN) classification for image processing, establishes a classification algorithm(CA) based on NN, and uses the attribute reduction(AR) of RS as the front end of neural network to solve the dimensionality problem when processing image information. Improves the accuracy of image classification(IC). The experimental results show that, compared with other CA, the rough set neural network CA in this paper can reduce the dimension of image attributes and achieve higher classification accuracy in image processing.