Haar wavelet based reduced reference quality assessment technique for JPEG/JPEG2000 images

We propose a new Reduced Reference model for assessing the quality of JPEG and JPEG2000 images. We utilize the Haar Wavelet Decomposition as a tool to model the image information along with the distortions presents in JPEG or JPEG2000 images. One of the superiority of our method is the ability to independently assess the quality of JPEG and JPEG2000 images without using any additional discrimination method. Our method also incorporate a simple design to model the distortions in the image without using complex Human Visual System (HVS) model. Despite its simple design, our method has demonstrated an accurate and reliable quality metric. Experimental results using several categories of JPEG and JPEG2000 images using LIVE Image Database Release 2 will be presented to verify our metric. The results of our experiments show that the method has achieved high correlation with the subjective data in most of the categories.