骨骼化.docVIP

细化也叫骨骼化。一个图像的“骨架”，是指图像中央的骨骼部分。是描述图像几何及拓扑性质的重要特征之一。求一图像骨骼的过程通常称为对图像“细化”过程。对图像细化有助于形状特征和减少冗余的信息量[76]。细化一副图像X的过程应该满足两个条件： 在细化的过程中，X应该有规律的缩小； 在缩小的过程中，应当使X的连同性质保持不变。 下面是具体的细化算法： 一副图像中的一个3×3区域，如右图所示，对各点标记名称P1，P2，…P8，其中P1位于中心。如果P1=1（即黑点），下面四个条件如果同时满足，则删除P1（P1=0）。 ①2≤NZ（P1）≤6；②Z0（P1）=1；③P2*P4*P8=0或者Z0（P1）≠1；④P2*P4*P6=0或者Z0（P4）≠1； 对图像中的每个点重复这一步骤，直到所有点都不可删除为止。 Private Sub thin() Dim i, j, m, n, count, u, v As Integer Dim pixel As Long Dim modify As Boolean Dim condition1 As Boolean Dim condition2 As Boolean Dim condition3 As Boolean Dim condition4 As Boolean Dim condition5 As Boolean Dim condition6 As Boolean Dim neighbour(0 To 4, 0 To 4) As Integer For j = 0 To 156 For i = 0 To 156 Picture5.PSet (j, i), Picture1.Point(j, i) picture5=picture1 Next Next For m = 0 To 4 For n = 0 To 4 neighbour(m, n) = 0 赋初值 Next Next modify = True While (modify = True) modify = False For j = 2 To 154 For i = 2 To 154 condition1 = False condition2 = False condition3 = False condition4 = False condition5 = False condition6 = False pixel = Picture5.Point(j, i) picture5提取象素值 pixel = pixel And (HFF) 取最低位 If pixel = 0 Then count = 0 For m = 0 To 4 For n = 0 To 4 pixel = Picture5.Point((j + m - 2), (i + n - 2)) 取该点与相邻24个点的象素 pixel = pixel And (HFF) 取最低位 neighbour(m, n) = (255 - pixel) / 255 If m 0 And n 0 And m 4 And n 4 Then count = neighbour(m, n) + count End If Next n Next m count = count - neighbour(2, 2) the first condtion 2=NZ(P1)=6 If count 1 And count 7 Then condition1 = True Else: Exit For End If count = 0 the second condtion Z0(P1)=1 If neighbour(1,