韶关学院学报
韶關學院學報
소관학원학보
Journal of Shaoguan University(Social Science Edition)
2010年
3期
25~28
,共null页
图像压缩 小波变换 神经网络 峰值信噪比
圖像壓縮 小波變換 神經網絡 峰值信譟比
도상압축 소파변환 신경망락 봉치신조비
image compression; wavelet transform ;neural network; peak signal-to-noise ratio
针对图像压缩中压缩率与图像质量的折衷问题。综合利用小波变换和神经网络各自的优点,采用小波和神经网络的方法进行图像压缩.该算法先对图像进行小波分解,保留低频系数,然后将高频系数输入训练的网络进行矢量量化编码达到压缩的目的.最后根据保留的低频系数和还原的高频系数重构图像.
針對圖像壓縮中壓縮率與圖像質量的摺衷問題。綜閤利用小波變換和神經網絡各自的優點,採用小波和神經網絡的方法進行圖像壓縮.該算法先對圖像進行小波分解,保留低頻繫數,然後將高頻繫數輸入訓練的網絡進行矢量量化編碼達到壓縮的目的.最後根據保留的低頻繫數和還原的高頻繫數重構圖像.
침대도상압축중압축솔여도상질량적절충문제。종합이용소파변환화신경망락각자적우점,채용소파화신경망락적방법진행도상압축.해산법선대도상진행소파분해,보류저빈계수,연후장고빈계수수입훈련적망락진행시량양화편마체도압축적목적.최후근거보류적저빈계수화환원적고빈계수중구도상.
In view of the trade-offs between the compression ratio and image quality, image compression method is adapted in the light of their respective advantages of the wavelet and neural network. Firstly, this algorithm carries on wavelet decomposition for the image, then retain low frequency coefficients and input the high frequency coefficient to the training network, the compressed image can be got by the vector quantization code, finally, the restructured image can be obtained according to retention of low frequency coefficient and high frequency coefficient. The experimental result indicate that this method is reasonable and feasible.