In this paper, we address the problem of no-reference quality assessment for digital pictures corrupted with blur. We start with the generation of a large real image database containing pictures taken by human users in a variety of situations, and the conduction of subjective tests to generate the ground truth associated to those images. Based upon this ground truth, we select a number of high quality pictures and artificially degrade them with different intensities of simulated blur (gaussian and linear motion), totalling 6000 simulated blur images. We extensively evaluate the performance of state-of-the-art strategies for no-reference blur quantification in different blurring scenarios, and propose a paradigm for blur evaluation in which an effective method is pursued by combining several metrics and low-level image features. We test this paradigm by designing a no-reference quality assessment algorithm for blurred images which combines different metrics in a classifier based upon a neural network structure. Experimental results show that this leads to an improved performance that better reflects the images' ground truth. Finally, based upon the real image database, we show that the proposed method also outperforms other algorithms and metrics in realistic blur scenarios.

No-Reference Blur Assessment of Digital Pictures Based on Multifeature Classifiers

This report discusses methods for estimating linear motion blur. The blurred image is modeled as a convolution between the original image and an unknown point-spread function. The angle of motion blur is estimated using three different approaches. The first employs the cepstrum, the second a Gaussian filter, and the third the Radon transform. To estimate the extent of the motion blur, two different cepstral methods are employed. The accuracy of these methods is evaluated using artificially blurred images with varying degrees of noise added. Finally, the best angle and length estimates are combined with existing deconvolution methods to see how well the image is deblurred.

https://conservancy.umn.edu/bitstream/handle/11299/3685/2133-5.pdf;sequence=1

Blind Image Deconvolution Motion Blur Estimation

Prior knowledge of the noise present in a color image acquisition device is very important in estimating colorimetric values or in recovering the spectral reflectances of pixels of objects being imaged, since these values are greatly influenced by the noise. In this paper, a new model is proposed for the determination of the noise variance of a multispectral color image acquisition system and experimental results to demonstrate its accuracy are presented. It is demonstrated that the noise variance of an actual multispectral color image acquisition system computed by the proposal agrees fairly well with the variance which minimizes the mean-square error of the recovered reflectances by the Wiener filter. As an application of the proposal, it is shown that spectral reflectances of an art painting are recovered accurately by the use of sensor responses without prior knowledge of objects being imaged and noise present in an image acquisition system.

Recovery of spectral reflectances of objects being imaged without prior knowledge

Motivated by a number of potential applications for a high data-density barcode that can be easily photographed and decoded by mobile phones, this paper presents the design of a new colour high data-density two-dimensional barcode symbology. The new symbology is designed to exploit the low-pass characteristic of a camera phone channel, and encodes information in the spatial Discrete Cosine Transform domain. A water-filling process and a noise-shaping algorithm enhance performance, while a new fast acquisition method allows for rotational and size invariance. An outer Accumulate-Repeat-Accumulate code is employed, followed by an inner Reed Muller code with a rate varying according to spatial frequency. The final barcode data-density is 3.5 times greater than the leading symbology and has proven robust to various impairments imposed by camera phones.

/pdf/two-dimensional-barcodes-for-mobile-phones-1tyroegue1.pdf

Two-dimensional barcodes for mobile phones

Methods for estimating the point spread function (PSF) of a motion-blurred image are disclosed and claimed. In certain embodiments, the estimated PSF may be used to compensate for the blur caused by hand-shake without the use of an accelerometer or gyro. Edge spread functions may be extracted along different directions from straight edges in a blurred image and combined to find the PSF that best matches. In other embodiments, the blur response to edges of other forms may similarly be extracted, such as corners or circles, and combined to find the best matching PSF. The PSF may then be represented in a parametric form, where the parameters used are related to low-order polynomial coefficients of the angular velocity vx(t) and vy(t) as a function of time.

Estimation of point spread functions from motion-blurred images

A method is proposed which estimates the degree of motion blur caused by the shaking of a digital camera during the exposure time. A new domain called the peak-trace domain is introduced to clearly specify the signal dominant region (used for extracting the camera-shaking degree) and the noise dominant region (used for estimating the noise variance). The experimental results show that the proposed method offers an efficient way to precisely detect the camera shaking and to restore the noisy blurred image.

An Algorithm To Extract Camera-shaking Degree And Noise Variance In The Peak-trace Domain.

Tracking of Moving Objects Using Bayes Decision and Cluster Model

While acquiring the image, the shaking of the acquiring device or of the object seriously damages the acquired image. This phenomenon, which decreases the distinction of the image is called motion blur. In this paper, a newly defined function is introduced for finding the degree and the length of the motion blur. The domain of this function is called the Peak-trace domain. In the Peak-trace domain, the noise dominant region -- for calculating the noise variance -- and the signal dominant region -- for extracting the degree and the length of the motion blur -- are defined. Using the information of the Peak- trace in the signal dominant region, we can fastly find the direction of the motion blur with noise immunity. A new weighted least mean square method helps extracting the Peak- trace more precisely. After getting the direction of the motion blur, we can find the degree of the motion blur very fast using a one dimensional Cepstrum method. In our experiment, we could efficiently restore the damaged image using the information we got by the above mentioned method.© (1998) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Ji Woong Choi

Papers

An Algorithm To Extract Camera-shaking Degree And Noise Variance In The Peak-trace Domain.

Tracking of Moving Objects Using Bayes Decision and Cluster Model

Algorithm for estimating the degree of camera shaking and noise corruption