Universal blurriness measurement approach for digital imagery
AssigneeDialogic Corporation
InventorChunhua Chen
Life of this patent
Estimating the term…
Abstract
Systems and methods of performing quantitative measurements of image blur in digital images and digital image sequences that are computationally efficient, and that employ no reference information in the measurement process. Each of the image blur measurements is performed using a Markov Chain, where a gradient image array is calculated for a pixel array derived from a given digital image, and a transition probability matrix is constructed for the transition probabilities between adjacent elements in the gradient image array. The transition probability data contained in the transition probability matrix can be pooled or otherwise accumulated to obtain a quantitative measurement of image blur in the given digital image.
Claims21
A system for measuring image blur in a digital image, the digital image having luminance and chrominance components, the system comprising:
- at least one memory;
- and
- at least one processor operative to execute at least one computer program out of the at least one memory:
- to generate, from at least one of the luminance and chrominance components of the digital image, image pixel data in an image pixel array corresponding to the digital image, the image pixel array having a plurality of pixel locations for pixel values of the image pixel data;
- to calculate gradient image data in a gradient image array from the image pixel data, the gradient image data representing a plurality of states, the gradient image array having a plurality of state locations for state values of the gradient image data, each of the state values corresponding to a respective gradient value calculated in a specified direction for a respective pair of adjacent pixel locations in the image pixel array;
- to estimate a plurality of transition probabilities between respective pairs of state values for adjacent state locations disposed in the specified direction throughout the gradient image array;
- to pool data corresponding to the plurality of transition probabilities to generate a measurement of image blur in the digital image;
- and
- to generate an output indicative of the measurement of image blur in the digital image.
A method of measuring image blur in a digital image, the digital image having luminance and chrominance components, the method comprising the steps of:
- generating, by an image pixel array generator from at least one of the luminance and chrominance components of the digital image, image pixel data in an image pixel array corresponding to the digital image, the image pixel array having a plurality of pixel locations for pixel values of the image pixel data;
- calculating, by a gradient image array generator from the image pixel data, gradient image data in a gradient image array, the gradient image data representing a plurality of states, the gradient image array having a plurality of state locations for state values of the gradient image data, each of the state values corresponding to a respective gradient value calculated in a specified direction for a respective pair of adjacent pixel locations in the image pixel array;
- estimating, by a transition probability generator, a plurality of transition probabilities between respective pairs of state values for adjacent state locations disposed in the specified direction throughout the gradient image array;
- pooling, by a data pooling component, data corresponding to the plurality of transition probabilities to generate a measurement of image blur in the digital image;
- and
- generating, by an output generator, an output indicative of the measurement of image blur in the digital image.
A system for measuring image blur in a digital image, the digital image having luminance and chrominance components, the system comprising:
- at least one memory;
- and
- at least one processor operative to execute at least one computer program out of the at least one memory:
- to generate, from at least one of the luminance and chrominance components of the digital image, image pixel data in an image pixel array corresponding to the digital image, the image pixel array having a plurality of pixel locations for pixel values of the image pixel data;
- to calculate gradient image data in a gradient image array from the image pixel data, the gradient image data representing a plurality of states, the gradient image array having a plurality of state locations for state values of the gradient image data, each of the state values corresponding to a respective gradient value calculated in a specified direction for a respective pair of adjacent pixel locations in the image pixel array;
- to estimate a plurality of transition probabilities between respective pairs of state values for adjacent state locations disposed in the specified direction throughout the gradient image array, each respective pair of state values including a current state value and a next state value, each of the adjacent state locations including a first state location and a second state location, each of the plurality of transition probabilities being a ratio of a co-occurrence probability that, when the current state value is disposed in the first state location, the next state value is disposed in the second state location, and an occurrence probability that the current state value is disposed in the first state location;
- to pool data corresponding to the plurality of transition probabilities to generate a measurement of image blur in the digital image;
- and to generate an output indicative of the measurement of image blur in the digital image.
Dependencies are inferred from the claim text.