Effects of JPEG Compression on Vision Transformer Image Classification for Encryption-then-Compression Images

Abstract

This paper evaluates the effects of JPEG compression on image classification using the Vision Transformer (ViT). In recent years, many studies have been carried out to classify images in the encrypted domain for privacy preservation. Previously, the authors proposed an image classification method that encrypts both a trained ViT model and test images. Here, an encryption-then-compression system was employed to encrypt the test images, and the ViT model was preliminarily trained by plain images. The classification accuracy in the previous method was exactly equal to that without any encryption for the trained ViT model and test images. However, even though the encrypted test images can be compressible, the practical effects of JPEG, which is a typical lossy compression method, have not been investigated so far. In this paper, we extend our previous method by compressing the encrypted test images with JPEG and verify the classification accuracy for the compressed encrypted-images. Through our experiments, we confirm that the amount of data in the encrypted images can be significantly reduced by JPEG compression, while the classification accuracy of the compressed encrypted-images is highly preserved. For example, when the quality factor is set to 85, this paper shows that the classification accuracy can be maintained at over 98% with a more than 90% reduction in the amount of image data. Additionally, the effectiveness of JPEG compression is demonstrated through comparison with linear quantization. To the best of our knowledge, this is the first study to classify JPEG-compressed encrypted images without sacrificing high accuracy. Through our study, we have come to the conclusion that we can classify compressed encrypted-images without degradation to accuracy.

Encryption Method for JPEG Bitstreams for Partially Disclosing Visual Information

“…When using the JPEG compression method, the compressed data can maintain the JPEG format [1][2][3]. In addition, this type of encryption has various applications including privacy-preserving deep learning [25,26]. However, given JPEG images prior to encryption, the JPEG images must be decompressed before carrying out encryption.…”

High Imperceptible Data Hiding Method for AMBTC Compressed Images Based on Combination Theory

“…This is due to the fact that computing is simple and inexpensive. There are several widely used compression methods, including vector quantization (VQ) [6]- [8], block truncation coding (BTC) [9]- [12], AMBTC [13], and JPEG compression [14]- [16]. Beginning by Delp and Mitchell [9] in 1979 which proposed block truncation coding (BTC) for image compression.…”

High Imperceptible Data Hiding Method Based on Combination Theory for AMBTC Compressed Images

“…This is due to the fact that computing is simple and inexpensive. There are several widely used compression methods, including vector quantization (VQ) [9][10][11][12], block truncation coding (BTC) [13][14][15][16], AMBTC [17,18], and JPEG compression [19][20][21], beginning with Delp and Mitchell [13] in 1979, who proposed block truncation coding (BTC) for image compression, and then further developed in 1984 by Lema and Mitchell [17], whose method is known as absolute moment block truncation coding (AMBTC). AMBTC offers simpler computation than BTC and acquires better image quality.…”