Method and apparatus for obscuring features of an image
AssigneeEastman Kodak Company
InventorRonald S. Cok
Life of this patent
Estimating the term…
Abstract
A method and system that obscures image features in a region of interest designated by a user. An operator designates a special shaped region of interest and a computer convolves a specially designed kernel with the image data in the region. The kernel is designed to pull features outside the region into the region while combining the information within the region with the information pulled from outside. The kernel can be designed to correspond to the shape of the region for efficient computation. Once the region of interest is obscured the image is printed resulting in a print that does not include undesirable features.
Claims3
A method of obscuring information in an image, comprising:
- (a) obtaining a digital signal representative of the image;
- (a) designating a region of the image for obscuration;
- and
- (b) merging first image signal information from outside the region with second image signal information from inside the region by applying at least one convolution kernel to the region from an outside edge of the region to a center in discrete sequential layers, the result of each layer being merged with the next layer;
- wherein the at least one convolution kernel is rotationally symmetric about a normal to the surface of the region, and wherein the at least one convolution kernel comprises at least eight rotationally symmetric kernels.
A method of obscuring information in an image, comprising:
- (a) obtaining a digital signal representative of the image;
- (a) designating a region of the image for obscuration;
- and
- (b) merging first image signal information from outside the region with second image signal information from inside the region by applying at least one convolution kernel to the region from an outside edge of the region to a center in discrete sequential layers, the result of each layer being merged with the next layer;
- wherein the it least one convolution kernel is v-shaped.
Dependencies are inferred from the claim text.