发明名称 Optimizing an image filter
摘要 A system may receive filter coefficients defining a digital filter. The system may select a signal processing quality criterion which describes a transformation that can be derived from an image and further describes the reconstruction of the image that can be derived from the transformation. The system may determine a degree of optimization that quantifies the signal processing quality criterion with the received filter coefficients. The system may vary the filter coefficients to obtain varied filter coefficients. The system may establish the degree of optimization with the varied filter coefficients. The system may compare the determined degree of optimization with the established degree of optimization.
申请公布号 US9589206(B2) 申请公布日期 2017.03.07
申请号 US201314097933 申请日期 2013.12.05
申请人 HOCHSCHULE PFORZHEIM 发明人 Greiner Thomas;Gehrke Tobias
分类号 G06K9/40;G06K9/36;G06K9/62;G06T3/00;H04N19/60;H04N19/635 主分类号 G06K9/40
代理机构 Harrity & Harrity, LLP 代理人 Harrity & Harrity, LLP
主权项 1. A computer-implemented method for optimizing a digital filter, the method comprising: receiving, by a processor, filter coefficients defining the digital filter, the digital filter being an analysis filter that is applicable to an image as part of a directional transform to derive a transformation of the image, orthe digital filter being a synthesis filter that is applicable to the transformation to derive a reconstruction of the image from the transformation,the filter coefficients defining a plurality of partial analysis filters; performing, by the processor, an expansion of products of matrices of the plurality of partial analysis filters to obtain complete analysis filters suitable for filtering the image; combining, by the processor, the complete analysis filters into an equivalent analysis filter; selecting, by the processor, a signal processing quality criterion that describes the transformation to be derived from the image and further describes the reconstruction of the image to be derived from the transformation; calculating, by the processor, a similarity function describing a degree of difficulty of deriving the transformation of the image; applying, by the processor, the equivalent analysis filter to the similarity function in order to determine a transformation error in the transformation of the image; determining, by the processor and based on the transformation error, a first degree of optimization that quantifies the signal processing quality criterion with the filter coefficients; varying, by the processor, the filter coefficients to obtain varied filter coefficients; establishing, by the processor, a second degree of optimization with the varied filter coefficients; comparing, by the processor, the first degree of optimization and the second degree of optimization; and determining, by the processor, preferred filter coefficients with a better degree of optimization based on comparing the first degree of optimization and the second degree of optimization, the preferred filter coefficients including the filter coefficients or the varied filter coefficients.
地址 Pforzheim DE