Image Quality Assessment of No Reference JPEG Compressed Images Using Various Spatial Domain Features

In today's world of technological advancement higher data rate and data transmission with minimal memory requirement is gaining importance. At the same time it also important to preserve quality of data to be transferred from various types of distortions and to assess the quality at the receiving end. Hence perceptual image quality assessment is becoming more popular. This paper introduces a new approach in domain of no reference image quality assessment of jpeg compressed images using spatial features. Considering the fact that pixel distortion, blurring and edge information are important aspects as far as distortions are concern. A novel approach is presented in this paper to address the distortion categories. The results are obtained in conjunction with the subjective image quality assessment of train and test image categories. Also results are compared with full reference image quality assessment technique and parameters of full reference quality assessment. To achieve better accuracy algorithm is tested on LIVE Texas' image database and our own developed database. Results are found to be well correlating to the various parameters considered for the comparison purpose.