Joint Power and Contrast Shrinking in RGB Images with Exponential Smoothing

In this study, an exponentially weighted moving average (EWMA) filter is employed to transform the intensity values of RGB colour images. The processing is applied at the YCbCr colour system and targets only the component Y. The rationale behind the selection of temporal filtering with exponential weighting is to generate images with lower power simply and efficiently. This filter gives lower importance to the previous intensity of each pixel, while higher weight is given to the input, which here captures the overall brightness of the image (not an individual brightness level per pixel). Compared to the Concurrent Brightness & Contrast Scaling (CBCS) linear mapping algorithm, the exponential filter can produce lower power images, forcing higher histogram shrinkage but also high perceptual quality. The structural similarity index (SSIM) and power decay rate metrics are evaluated to verify their performance and empirically identify the optimal values for the involved parameters.