In the No-Reference Image Quality Assessment, it is important to play a role in accurate positioning accuracy of the algorithm on image quality. Based on the current application of no reference image evaluation algorithm of some mainstream in a public database of research, including based on edge detection, based on energy distribution and based Pixel semaphore. Experiments show that the test accuracy of the algorithm based on edge detection can reach 95% in‘TID2013’data set and 91% in the traffic database and the speed of each picture is 0.002s, which is obviously better than the evaluation algorithm based on energy distribution and the evaluation algorithm based Pixel semaphore. It can be used as the basis of image evaluation for traffic monitoring system.
The problem of image matching and target tracking based on singular value decomposition (SVD) is discussed. The SVD
has robust performance that is invariant to image disturbance and it makes the singular value credible to represent the
image as an algebraic feature. A template-updating strategy is proposed to update the current template based on the scale
invariant character of the singular value vector. The updated template that contains the accurate target is adaptively
acquired according to the singular value's scale invariance. Experiments are performed on a large test set and the results
show that the proposed strategy is practical and efficient in target tracking.
The paper presents a new algorithm for image de-striping. A kind of adaptive frequency filter is constructed based on two-dimension fast Fourier Transform in the proposed algorithm. To construct the adaptive filter, a stripe frequency band (SFB) and an accumulation distribution function (ADF) are defined in the fourier power spectrum. Comparing to the traditional de-striping methods, the algorithm does not take on abundant computation load and need no manual participating to outline the stripe band in frequency domain. Through the evaluation of visual effect and SNR/PSNR comparison, the experiment results demonstrate the proposed algorithm can remove the strips effectively while contain the original image information perfectly.