JPEG [1] is a widely used image compression format. It has a great economic value for making the image file smaller but guarantee the image visual quality. Guetzli [2] is applied to compress and obtain images conforming to the JPEG standard. It evaluates the quality of the image through Butterugli [3], compresses the JPEG image by replacing the quantization table and filtering the DCT coefficients. The advantage of Guetzli is that it uses Butterugli to restrict and supervise the image optimization process, so that the file is reduced while the image quality is strictly guaranteed. However, the disadvantage of Guetzli is the time cost of its iterative process and the considerable space complexity during the calculation. To solve this problem, two methods are proposed to speed up the Guetzli calculation process. First, tighten the preliminary replacement process of DCT coefficients. Second, speed up the backfill process of the DCT coefficients. And the combination of these two methods can be used to speed up the compression process. Our method can effectively speed up the Guetzli compression process without lead into extra distortion of images. As a result, on the premise of ensuring quality, 73% of the iteration time savings was realized at the cost of 3.8% file size increase.
[1]
Jyrki Alakuijala,et al.
Guetzli: Perceptually Guided JPEG Encoder
,
2017,
ArXiv.
[2]
K WallaceGregory.
The JPEG still picture compression standard
,
1991
.
[3]
Mohamed Cheriet,et al.
Mean Deviation Similarity Index: Efficient and Reliable Full-Reference Image Quality Evaluator
,
2016,
IEEE Access.
[4]
Eero P. Simoncelli,et al.
Image quality assessment: from error visibility to structural similarity
,
2004,
IEEE Transactions on Image Processing.
[5]
David Zhang,et al.
FSIM: A Feature Similarity Index for Image Quality Assessment
,
2011,
IEEE Transactions on Image Processing.