应用科技
應用科技
응용과기
YING YONG KE JI
2014年
6期
22-26
,共5页
张琦%李鸿林%杨大伟%李松江
張琦%李鴻林%楊大偉%李鬆江
장기%리홍림%양대위%리송강
无参考%图像质量评价%人类视觉系统%局部方差%JPEG压缩图像
無參攷%圖像質量評價%人類視覺繫統%跼部方差%JPEG壓縮圖像
무삼고%도상질량평개%인류시각계통%국부방차%JPEG압축도상
no reference%image quality assessment%human vision system%local variance%JPEG compressed images
针对传统JPEG压缩图像质量评价算法存在高计算复杂度的问题,通过改进经典的JPEG压缩图像质量评价算法,以无参考为基础,提出一种快速有效的质量评价算法。该算法将人眼视觉系统特性引入到图像质量评价体系中,利用局部方差选取人眼感兴趣图像块代替整体图像,并只针对感兴趣图像块做特征提取处理,计算得出3个图像的特征值,最后将所有特征值整合为一个评价值,获取原始整体图像的客观质量评价参数。仿真测试结果表明,与传统整体图像JPEG压缩图像评价算法相比,该算法皮尔逊相关系数提高0.01,与主观评价结果更为一致;运算速度提高一倍,降低了运算复杂度。
針對傳統JPEG壓縮圖像質量評價算法存在高計算複雜度的問題,通過改進經典的JPEG壓縮圖像質量評價算法,以無參攷為基礎,提齣一種快速有效的質量評價算法。該算法將人眼視覺繫統特性引入到圖像質量評價體繫中,利用跼部方差選取人眼感興趣圖像塊代替整體圖像,併隻針對感興趣圖像塊做特徵提取處理,計算得齣3箇圖像的特徵值,最後將所有特徵值整閤為一箇評價值,穫取原始整體圖像的客觀質量評價參數。倣真測試結果錶明,與傳統整體圖像JPEG壓縮圖像評價算法相比,該算法皮爾遜相關繫數提高0.01,與主觀評價結果更為一緻;運算速度提高一倍,降低瞭運算複雜度。
침대전통JPEG압축도상질량평개산법존재고계산복잡도적문제,통과개진경전적JPEG압축도상질량평개산법,이무삼고위기출,제출일충쾌속유효적질량평개산법。해산법장인안시각계통특성인입도도상질량평개체계중,이용국부방차선취인안감흥취도상괴대체정체도상,병지침대감흥취도상괴주특정제취처리,계산득출3개도상적특정치,최후장소유특정치정합위일개평개치,획취원시정체도상적객관질량평개삼수。방진측시결과표명,여전통정체도상JPEG압축도상평개산법상비,해산법피이손상관계수제고0.01,여주관평개결과경위일치;운산속도제고일배,강저료운산복잡도。
A fast and effective quality assessment algorithm based on no?reference was proposed to solve the problem of high computational cost in typical quality assessment algorithm for JPEG compressed images. The proposed algo?rithm introduced the characteristics of human visual system into the image quality assessment system, using the lo?cal variance to select the image blocks interested by human being instead of selecting the entire image. It conducted feature extraction only for the interested image blocks. Three eigenvalues of the image were calculated, and at last, all eigenvalues were integrated as one assessment value to obtain the objective quality evaluation parameters of the original entire image. The simulation results show that compared to the traditional method, the proposed algorithm is more consistent with the subjective evaluation results with the Pearson correlation coefficient increased 0.01 and less complex with half running time.