Thumbnail Image

Theses

Browse

20161
Reset filters
Subject: Electronics and Communication Engineering × Author: Agarwal, Saitu × Start date: 2016 × Type: Thesis ×
Reset filters

Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    ThesisItemOpen Access
    Reconstruction of motion blurred image using wavelet transform and neural committee machine
    (G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand), 2016-08) Agarwal, Saitu; Mathur, Sanjay
    Image Restoration is the process of obtaining the original image by using the knowledge of the degrading factors. Degradation comes in many forms such as blur, noise, and camera shake. One of the disadvantages of existing method is that some methods make restrictive assumptions on the Point Spread Function or the true image that limits the algorithm's portability to different applications in conventional approach; deblurring filters are applied on the degraded images without the knowledge of blur and its effectiveness. The proposed algorithm is used for the reconstruction of motion blurred images. Two factors which mainly affect an image are length of blur and angle of blur. In this thesis, concept of Wavelet decomposition and neural committee machines are applied for restoring problems in which images are degraded by blur function and corrupted by noise. In the first step of reconstruction, blurred image is decomposed using wavelet transform based on multi resolution analysis and then correlation analysis is performed to reduce the dimensionality of image pattern space and in the next step a multi layer feed forward neural network based on ensemble averaging technique is used to estimate the parameter which causes blurring. The proposed methodology uses highly non linear back propagation neuron for image restoration to get a high quality of restored image. The estimated parameters are used to calculate the value of degradation factor corresponding to blurred image and then deconvolution is performed to get the original image. Results show that the images restored using the proposed method have mean square error in the range of 0.001-.005 and peak signal to noise ratio in the range of 69dB-75dB.