Low-Complexity Transcoding of JPEG Images With Near-Optimal Quality Using a Predictive Quality Factor and Scaling Parameters

Abstract

A common transcoding operation consists of reducing the file size of a JPEG image to meet bandwidth or device constraints. This can be achieved by reducing its quality factor (QF) or reducing its resolution, or both. In this paper, using the Structural SIMilarity (SSIM) index as the quality metric, we present a system capable of estimating the QF and scaling parameters to achieve optimal quality while meeting a device's constraints. We then propose a novel low-complexity JPEG transcoding system which delivers near-optimal quality. The system is capable of predicting the best combination of QF and scaling parameters for a wide range of device constraints and viewing conditions. Although its computational complexity is an order of magnitude smaller than the system providing optimal quality, the proposed system yields quality results very similar to those of the optimal system.

Quality-aware predictor-based adaptation of still images for the multimedia messaging service

“…We have shown in previous work that strategies involving both adaptation of JPEG compression parameters and scaling produce significantly better results than using either method alone [13]. In this work, we propose to extend this approach to the explicit optimization of compression parameters and scaling over a complete, multipart, image-only message, in order to achieve an adaptation that not only satisfies the constraints of the receiving device, but also maximizes overall perceived quality, with the expectation that the user experience will be maximized.…”

“…Since the images m i and T (m i , t i ) may differ in resolution, that is, whenever z = 1, the image T (m i , t i ) is rescaled to the original resolution R(m i ) before comparison, estimating the viewing condition of the sender. Other strategies could be considered; for example, comparing at the scale reproducing the viewing condition of the receiver, or at an intermediate scale between sender and receiver conditions [13]. However, it seems realistic to use the sender's viewing conditions for the comparison, in order to estimate degradation in some absolute sense.…”

“…The quality factor, as defined by the IJG, is an integer that takes values from 0 to 100 inclusive, but the scaling factor 0 < z i 1 can take an infinite number of values. We solved this problem in previous work [12,13,33] by constraining the two transcoding parameters to take at most ten distinct values, that is, the quality factors were limited to the set {10, 20, . .…”

“…To this end, we have proposed predictors and systems in previous work designed to adapt images and messages [12,13,33]. These predictors have been exploited in [21,22] to develop a dynamic content adaptation framework applied to collaborative mobile presentations (e.g.…”

“…To speed up optimization significantly, we propose to use predictors that yield a predicted file size and a predicted perceived image quality given particular adaptation parameters. In this work, we use a predictor that we presented in previous work [13], JQSP (for JPEG Quality and Size Predictor), which will serve as a "real life" prediction algorithm that exhibits moderate prediction errors, and compare it to oracular predictors that "predict" without error the resulting file size and image quality We show that the proposed optimization framework is resilient and shows graceful degradation in the presence of increased prediction error. We further show that our proposed optimization framework not only yields better perceived quality than approaches found in, and inspired by, prior art, but also that the transcoding times are significantly lower.…”