标题: matlab中的conv2卷积运算的三种形式 [打印本页] 作者: thinkfunny 时间: 2020-1-14 10:27 标题: matlab中的conv2卷积运算的三种形式 5 B+ q: C9 A( o
6 e5 c! T' P% A7 h e: B; ^5 o& R
matlab中的conv2是用于对二维数据进行卷积运算，有三个参数可供选择，下面是help content of conv2 l0 _2 P% [4 j( T( x; p9 t
. A) `7 y4 t9 u$ O+ f
conv2 Two dimensional convolution. # P: t4 q$ a& x* @% p k/ W0 }4 @, w! e! S# N% H$ k
C = conv2(A, B) performs the 2-D convolution of matrices A and B., c8 g! t! P/ u3 m x3 `" ]
If [ma,na] = size(A), [mb,nb] = size(B), and [mc,nc] = size(C), then1 ^; u4 I4 _5 P
mc = max([ma+mb-1,ma,mb]) and nc = max([na+nb-1,na,nb]).& j7 S; i3 ^. T) }- n, V
# a2 T8 a$ l* Y& W& s: B C = conv2(H1, H2, A) first convolves each column of A with the vector- \% h7 k( e# T5 }) P- A9 n
H1 and then convolves each row of the result with the vector H2. If ! i$ d7 x" ~, z n1 = length(H1), n2 = length(H2), and [mc,nc] = size(C) then * `! q3 b' p% `8 J& _) b mc = max([ma+n1-1,ma,n1]) and nc = max([na+n2-1,na,n2]). / }- v8 A- j4 ^: `$ E conv2(H1, H2, A) is equivalent to conv2(H1(：)*H2(：).', A) up to 9 s# N0 `: `9 F8 C% {: _5 T round-off. ' Q7 l a; |8 ^" J ( M% }! n) o. p5 m0 r9 y% D C = conv2(..., SHAPE) returns a subsection of the 2-D , l, q. ^# `* ^1 C3 G convolution with size specified by SHAPE:9 R+ J5 j* K, q
'full' - (default) returns the full 2-D convolution, 9 `& N- W. L' e* @4 J% Y 'same' - returns the central part of the convolution D3 I. G4 Z* o ? that is the same size as A.! R: a! c$ p& d9 {. \3 N
'valid' - returns only those parts of the convolution$ m, l% |6 D* o8 E8 l |- j
that are computed without the zero-padded edges. - i- t# h/ }, P8 A' K9 B9 V size(C) = max([ma-max(0,mb-1),na-max(0,nb-1)],0). , C. A: q6 M% V3 M4 G% f7 ]! G) O9 q. o* N$ I- i$ N( Y5 l$ F$ _
, {7 `5 c" R% }9 K关于full, same以及valid三种参数的区别，如下面的实例所示：- c5 e. V$ m: r, L3 O# H9 k; ~/ y
7 [. Y' Y# G& g0 O( v
full4 z5 b+ y& k t* ^- R
