Digital Image Processing
Camera models and machine perception
Camera models and machine perception
Image Processing, Analysis, and Machine Vision
Image Processing, Analysis, and Machine Vision
Adaptive wavelet thresholding for image denoising and compression
IEEE Transactions on Image Processing
Hi-index | 0.00 |
Images and video captured in real world situations generally have distorted digital pixel values. A variety of situations can cause these image degradations: sensor motion, environmental conditions and random noise. A crucial procedure in computer vision is the assessment and quantification of digital image quality. A numerical score for describing image quality is useful for a number of applications, some of which include improving the performance of an image acquisition system and adaptive algorithms. We present an intuitive quality metric for characterizing the amount of blur in an image, through blind image assessment, using the Haar discrete wavelet transform. Thus, the method does not require a reference image or any prior information. The novelty of our method lies in processing the image derivative using the discrete wavelet transform rather than directly processing image intensity values as is traditionally done. We present late breaking results and analysis for a small set of data. The proposed method shows promise for a large number of avenues such as realtime blur level assessment and image depth of focus estimation.