Ceramics first emerged in ancient China, and the development of ceramic art has a long history. In today's digital age, how to use the power of science and technology to conduct research on ceramic art is a problem worthy of attention. Digital Image Processing (DIP for short) is a method of using computer to perform graphics calculation, which has the characteristics of high efficiency and intelligence. It aims to use DIP technology to study ceramic art. In this regard, it has proposed to use convolutional neural network (CNN for short) to extract image features, identify and detect ceramic artworks, and has used mean filtering and median filtering to optimize the noise reduction of the identified images, so that the obtained image Higher quality. In the simulation test, it has selected 50 ceramic products for image recognition and detection, and has divided them into two groups for analysis. The results have showed that the inverse histograms of the two sets of images could reflect the defective parts of the ceramic products. Based on this, the lowest accuracy rate of CNN recognition was over 85%, and the highest accuracy rate was 94%. In the first group of images, the SNRs obtained by mean filtering and median filtering are the lowest of 7.7, 6.5, the highest are 8.7, 8.3, and the average SNR is about 8.0, 7.2; the SNR obtained by the two filtering methods of the second group of images is around 6.5 to 8.0, and the SNR of the mean filtering is slightly higher. Therefore, some practical results have been achieved in the research of ceramic art using DIP technology.