发明名称 IMAGE PROCESSING DEVICE
摘要 Provided is an image processing device capable of reproducing an original thin linear image region from a linear image region, which has become thick due to upscaling processing for a low-resolution image. In an image processing device (100), an amplitude estimation unit (1) calculates a magnification a in a local region so that an input image signal becomes similar to a low-resolution pattern PtnL. A similarity calculation unit (2) outputs a similarity s, which has a great value, at a pixel position which is similar to the low-resolution pattern PtnL. An FIR filter unit (4) sets an FIR filter coefficient by a difference pattern PtnDiff, receives a product of the magnification a and the similarity s, and performs FIR filter processing, and thereby acquires a high frequency component ΔD. Then, when the high frequency component ΔD is added to an input image Din by an adder (5), an output image Dout having excellent definition is acquired.
申请公布号 US2016307293(A1) 申请公布日期 2016.10.20
申请号 US201515036868 申请日期 2015.01.13
申请人 SHARP KABUSHIKI KAISHA 发明人 MURAHASHI Yoshimitsu
分类号 G06T3/40;G06T5/00;G06K9/62 主分类号 G06T3/40
代理机构 代理人
主权项 1. An image processing device which sharpens an input image, comprising: an amplitude estimator which acquires a magnification a so that a difference between an adjusted low-resolution pattern data sequence acquired by multiplying a low-resolution pattern data sequence, which is acquired by executing predetermined processing with respect to a high-resolution pattern data sequence, by the magnification a and a pixel data sequence in a local region that is a predetermined range of the input image, which includes a pixel to be processed, is equal to or less than a predetermined value; a similarity calculator which sets a similarity value of the pixel to be processed to be a value indicating that a degree of similarity between the adjusted low-resolution pattern data sequence and the pixel data sequence is higher as the difference between the adjusted low-resolution pattern data sequence and the pixel data sequence is smaller in the local region; a multiplier which multiplies the magnification a and the similarity for each pixel to be processed; an FIR filter which executes FIR filter processing with respect to a multiplication value of the magnification a and the similarity, which is acquired by the multiplier, with a difference pattern data sequence acquired by subtracting the high-resolution pattern data sequence from the low-resolution pattern data sequence as an FIR filter coefficient, and thus the FIR filter acquires high frequency component data ΔD of the pixel to be processed, for each pixel to be processed; and an adder which adds pixel data of the pixel to be processed and the high frequency component data ΔD of the pixel to be processed, which is acquired by the FIR filter, for each pixel to be processed.
地址 Osaka JP