计算机技术与发展
計算機技術與髮展
계산궤기술여발전
COMPUTER TECHNOLOGY AND DEVELOPMENT
2013年
1期
63-66
,共4页
梯度%复数矩阵%图像质量评价%奇异值分解%人类视觉系统
梯度%複數矩陣%圖像質量評價%奇異值分解%人類視覺繫統
제도%복수구진%도상질량평개%기이치분해%인류시각계통
gradient%complex matrix%image quality assessment%singular value decomposition%human visual system
将灰度图像的梯度信息作为表征图像结构信息的一个重要特征,进一步提高了图像质量客观评价结果与主观评价结果的一致程度,增强了人眼敏感的图像边缘信息.对梯度复数矩阵进行奇异值分解,计算得到相应的奇异值特征向量,通过计算参考图像与降质图像对应分块复数矩阵奇异值特征向量的相似程度得到了降质图像的量化质量评价结果.实验表明,梯度信息更能突出图像的结构特征,所提方法优于传统 MSE 及 MSVD 方法,与人眼视觉特性的一致性较好.
將灰度圖像的梯度信息作為錶徵圖像結構信息的一箇重要特徵,進一步提高瞭圖像質量客觀評價結果與主觀評價結果的一緻程度,增彊瞭人眼敏感的圖像邊緣信息.對梯度複數矩陣進行奇異值分解,計算得到相應的奇異值特徵嚮量,通過計算參攷圖像與降質圖像對應分塊複數矩陣奇異值特徵嚮量的相似程度得到瞭降質圖像的量化質量評價結果.實驗錶明,梯度信息更能突齣圖像的結構特徵,所提方法優于傳統 MSE 及 MSVD 方法,與人眼視覺特性的一緻性較好.
장회도도상적제도신식작위표정도상결구신식적일개중요특정,진일보제고료도상질량객관평개결과여주관평개결과적일치정도,증강료인안민감적도상변연신식.대제도복수구진진행기이치분해,계산득도상응적기이치특정향량,통과계산삼고도상여강질도상대응분괴복수구진기이치특정향량적상사정도득도료강질도상적양화질량평개결과.실험표명,제도신식경능돌출도상적결구특정,소제방법우우전통 MSE 급 MSVD 방법,여인안시각특성적일치성교호.
Structural information was described by gradient to improve the consistency of objective image assessment method with subjec-tive method. The effect of the edge structure in the image quality was emphasized accordingly. Singular value decomposition was per-formed on gradient distribution complex matrix. The similarity of the singular vectors of the reference image and the distorted image was used to measure the structural similarity of the two images. Then the numerical quality assessment results were achieved. Result from ex-periments shows that the gradient distribution can emphasize the edge information. The proposed method is better consistent with human visual system characteristics than MSE and MSVD method based on simple pixel value distribution.
