摘要
在图象压缩、重构等处理中，二进正交小波变换是一个非常有用的工具，其函数形式为{2^k/2Wn(2^kt-j):n∈N;k,j∈z}(t是实数）；在计算过程中，由于数据量太大，为了减少计算量应选择运算较快的算法，在计算机中浮点数比整数运算速度慢，若把t浮点数表示为形如A2^k+1(A是整数）的形式时，则2^kt-j为整数的形式，可以大大提高计算速度；在进行上述变换时，在考虑计算机字长p的限制下，为使在加、减、乘运算时，运算后的数据误差满足小于给定的ε、k、A如何取值，并得到了ε与p的一个近似关系式。
The orthonoimal dyadic wavelet is a powerful tool in image compress and reconstruction, its functionform is (t is real number) , Due to much data when computing, the faster way of computation should be employed to reduce the computing quantity. In computer the floating point number is at lower speed than integer number, if transform t to form of A/(2k+1), the 2kt - j must be integer ,speed will be higher. While transform the real numbert into the form of A/(2k+1) how to make sure the k , A incondition of bytes of computer and when sum, sub, product the numbers so that the error must be less than the given ε , We get a similar equation of ε and p .
出处
《郑州工业高等专科学校学报》
2003年第1期37-39,共3页
Journal of Zhengzhou Polytechnic Institute
