Image Signal Normalization via Polynomial Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for image signal normalization in video and still image quality measurement fail to accurately correct for brightness, contrast, color space, and gamma correction errors introduced during encoding, transmission, and decoding, particularly when different color space representations are used by various elements in the transmission chain.
Innovation Solution
A third-order polynomial mapping is used to normalize the effects of brightness, contrast, and gamma correction errors between reference and decoded image signals, with coefficients optimized to minimize the error between the two signals, employing singular value decomposition to generate a non-linear transformation vector for luminance and chrominance components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a comparison between reference and decoded signals is used for quality measurement, then quality assessment can be performed, but brightness, contrast, color space and gamma modifications introduced during encoding, transmission and decoding cause measurement errors
Solution Approach 1:
The patent applies preliminary normalization to the decoded signal before quality measurement. A transformation vector is computed from the reference signal and applied to the decoded signal in advance, correcting brightness, contrast, color space and gamma modifications before the quality assessment comparison, thereby eliminating measurement errors caused by these modifications
Solution Approach 2:
The patent changes the parameters of the decoded signal by applying a transformation vector that adjusts brightness (offset), contrast (gain), color space (color gain/offset), and gamma (power law) parameters. This transforms the decoded signal parameters to match the reference signal parameters, enabling accurate quality measurement
2Adaptability or versatility
If display settings such as brightness, contrast and gamma correction are applied, then the image can be adapted to different display devices, but these settings modify the received signal and prevent accurate quality measurement
Solution Approach 1:
The patent performs preliminary normalization of the decoded signal using a transformation vector derived from the reference signal, correcting display-related modifications (brightness, contrast, gamma, color space) before quality assessment. This allows quality measurement to be performed independently of the receiver's display settings while still allowing display adaptation
3Ease of manufacture
If linear transformations are used to correct brightness and contrast, then these corrections can be applied, but gamma correction requires power-law transformation which complicates the normalization process
Solution Approach 1:
The patent merges multiple correction operations (brightness offset, contrast gain, color space transformation, and gamma power-law correction) into a single unified transformation vector application process. This combines several linear and non-linear transformations into one comprehensive normalization step, reducing the overall process complexity while maintaining correction effectiveness
Data Source
AI summary
The present invention relates to an apparatus and method for normalization of an image signal, particularly for use in perceptual video and still image quality measurement when the quality measurement is achieved by making a comparison between a reference signal and a decoded signal. The invention provides a method for normalizing a decoded image signal in relation to a reference image signal to generate a normalized image signal comprising the steps of: generating a non-linear transformation vector in dependence upon at least one component of the decoded image signal and at least one component of the reference image signal; and applying said non-linear transformation vector to at least one component of the decoded image signal to generate a normalized image signal. Methods and apparatus for perceptual video and still image quality measurement using said method are also provided.