ESTIMATING DAMAGED PIXEL USING NEIGHBORHOOD INFORMATION FOR DIGITAL IMAGE INPAINTING

Estimatingthe value of deadpixels in a damaged image isoften desired in image processing. This technique is calledas Image In painting. Inthis paper,two Image In paintingtechniquesbased on neighborhood information are compared.In technique one a kernel function of dynamic weights is calculated based on the direction between known pixel and unknown pixel and the Euclidian distance between them. Whilein technique two the a kernel function is calculated based on gradient of pixels in a neighborhood. The image is unpaintedin both the techniquesby convolving the damaged region with the kernel function. Thetwo techniques are testedonimages and theexperimentalresults confirmthe effectiveness of the algorithm