Textile and clothing art patterns are the basis of clothing design, and modern clothing design is moving towards intelligent development. In this study, the VGG19 network is first used to construct a style migration model based on attention mechanisms to realize image style conversion for the intelligent generation of textile and clothing art. The keying algorithm is optimized using a convolutional network to segment textile and clothing art images and produce more transparent images. In using textile and clothing art patterns, the signal-to-noise ratio of the style migration model in this paper is between 9.39 and 25.53 dB, and the structural similarities are all above 0.7. The average score of its visual effect is 4.2, and the scores of each dimension of pattern generation and the scores of the five aspects of consumers are all above 4. The style migration model constructed in this paper has excellent performance in the use of textile and clothing art pattern design, which provides an effective way for the use of style migration network in the field of textile and clothing art patterns, and improves the efficiency of textile and clothing art pattern design and creation.