编成放大图片像素的方法.docx

1、编成放大图片像素的方法要处理一个图像，首先要获得该图像的像素值，而VB本身提供的PICTURE控件虽然可以打开很多类型的图片，但是它提供的那个POINT方法读取像素实在是太慢。而使用GetPixel这个API的速度也快不到哪里去，因为PIONT方法本身就是对于GetPixel的一个包装。在VB中要快速获取一幅在PICTURE中打开的图像比较快速的方法是使用DIB方法，当然还有DDB方法，不过使用DDB方法还需要考虑不同颜色深度的图像的分别处理，在程序的实现上要相对复杂，而使用DIB方法则不必，并且在处理速度上比DDB方法也慢的有限。 过程一：获得一个在PICTURE控件中打开的图像的所有像素。

2、 Public Sub DibGet(ByVal IdSource As Long, XBegin As Long, ByVal YBegin As Long, ByVal XEnd As Long, ByVal YEnd As Long)Dim iBitmap As LongDim iDC As LongDim I As LongDim Dim W As LongDim H As LongOn Error GoTo ErrLineDone = FalseTimeGet = timeGetTimeInPutWid = XEnd - XBeginInPutHei = YEnd - YBeginW

3、 = InPutWid + 1H = InPutHei + 1I = (Bits 8) - 1ReDim ColVal(I, InPutWid, InPutHei)With bi24BitInfo.bmiHeader.biBitCount = Bits.biCompression = 0&.biPlanes = 1.biSize = Len(bi24BitInfo.bmiHeader).biWidth = W.biHeight = HEnd WithiBitmap = GetCurrentObject(IdSource, 7&)GetDIBits IdSource, iBitmap, 0&,

4、H, ColVal(0, 0, 0), bi24BitInfo, 0& DeleteObject iBitmapDone = TrueTimeGet = timeGetTime - TimeGetExit SubErrLine:MsgBox 错误号： & Err.Number & ： & Err.DescriptionEnd Sub在这个过程中所用到的只是一些参数的设定和API的调用，不涉及算法。过程二：图像输出的过程：Public Sub DIBPut(ByVal IdDestination As Long)Dim W As LongDim H As LongOn Error GoTo Er

5、rLineDone = FalseTimePut = timeGetTimeW = OutPutWid + 1H = OutPutHei + 1With bi24BitInfo.bmiHeader.biWidth = W.biHeight = HLineBytes = (W * Bits + 31) And &HFFFFFFE0) 8.biSizeImage = LineBytes * HEnd WithSetDIBitsToDevice IdDestination, 0, 0, W, H, 0, 0, 0, H, ColOut(0, 0, 0), bi24BitInfo.bmiHeader,

6、 0 Done = TrueTimePut = timeGetTime - TimePutExit SubErrLine:MsgBox Err.DescriptionEnd Sub下面解释一下在过程中到的全局变量和数据结构，以及API的定义。API定义：删除一个DCPrivate Declare Function DeleteDC Lib gdi32 (ByVal hdc As Long) As Long删除一个对象Private Declare Function DeleteObject Lib gdi32 (ByVal hObject As Long) As Long选择当前对象Priva

7、te Declare Function GetCurrentObject Lib gdi32 (ByVal hdc As Long, ByVal uObjectType As Long) As Long获取DIBPrivate Declare Function GetDIBits Lib gdi32 (ByVal aHDC As Long, ByVal hBitmap As Long, ByVal nStartScan As Long, ByVal nNumScans As Long, lpBits As Any, lpBI As BitMapInfo, ByVal wUsage As Lon

8、g) As Long获取系统时间Private Declare Function timeGetTime Lib winmm.dll () As Long数据结构定义：Private Type BitMapInfoHeader 文件信息头BITMAPINFOHEADERbiSize As Long biWidth As Long biHeight As Long biPlanes As Integer biBitCount As Integer biCompression As Long biSizeImage As Long biXPelsPerMeter As Long biYPelsPe

9、rMeter As Long biClrUsed As Long biClrImportant As Long End TypePrivate Type RGBQuadrgbBlue As BytergbGreen As BytergbRed As BytergbReserved As ByteEnd TypePrivate Type BitMapInfobmiHeader As BitMapInfoHeaderbmiColors As RGBQuadEnd Type这三个数据结构都是在DIB中不可缺少的。我们不必深究，只是按照顺序复制粘贴直接使用就是了。过程中用到的全局变量：Private

10、Const Bits As Long = 32 颜色深度，这里把所有图像都按照32位来处理Public Done As Boolean 用于标记一个过程是否结束Public TimeGet As Long 用于记录输入过程处理所花费的时间Public TimePut As Long 用于记录输出过程处理所花费的时间Dim ColVal() As Byte 用于存放从DIB输入的像素值Dim ColOut() As Byte 用于存放向DIB输出的像素值Dim InPutHei As Long 用于记录输入图像的高度Dim InPutWid As Long 用于记录输入图像的宽度Dim bi24

11、BitInfo As BitMapInfo 定义BMP信息可以看出，我在输入和输出中使用了两个不同的动态数组ColVal()和ColOut()，这么做是有道理的，因为我们不只是为了输入和输出图像，中间还要对像素进行处理。包括图像缩放、色彩调整、锐化、柔化等等处理，使用两个不同的数组来分别存放数据更有利于程序的实现。有些性急的朋友说不定已经把程序贴到工程里试用了，可是会发现根本不能输出图像。这是因为当你用DIBGET获得的图像还在ColVal() 中呢，需要把它们放到ColOut()这个数组中去，DIBPUT这个过程才能起作用。这里再给出一个用于数组整体移动数据的过程：Public Sub Co

12、pyData(ByVal W As Long, ByVal H As Long)Dim Length As LongDim I As LongDim L As LongI = Bits 8L = I - 1Length = (W + 1&) * (H + 1&) * IReDim ColOut(L, W, H)CopyMemory ColOut(0, 0, 0), ColVal(0, 0, 0), LengthEnd subAPI定义：Private Declare Sub CopyMemory Lib kernel32 Alias RtlMoveMemory (pDest As Any, p

13、Src As Any, ByVal ByteLen As Long)这时，我们就可以来试一下效果了：把你的显示器调到32位色。将前面的所有API和变量定义全部贴到一个新建的模块里新建一个窗体，加两个PICTURE控件：pictrue1 ,picture2 一个按钮command1在pictrue1中加载一个图片在command1中写如下代码：sub command1_click()With picture1.ScaleMode=3.BorderStyle=0DibGet .hdc,0,0,.scalewidth,.scaleheightEnd WithCopyData InPutHei ,In

14、PutWidpicture2.AutoRedraw=TrueDibPut picture2.hdcpicture2.refreshend sub运行一下，按钮按下，pictreu1中的图片就立刻显示到了picture2中。这时，你可能会说，弄了这么半天就贴个图？用PaintPicture不是就可以了吗？不错，如果只是要贴个图，确实不用这么麻烦，可是，我们后面要说的图像处理部分将会用到前门得到的像素值。所以，这只是一个开始，我真正要讲的东西还在后面呢。请大家继续关注。前面讲到了二次线性插值的应用。这一篇来给大家讲一下关于锐化、柔化、扩散、雕刻这几个滤镜的实现。 一、锐化锐化的算法很简单，就是比较

15、相邻的几个像素，把当前像素加上和周围的像素的差就可以了。这里我给出一个示例：A B C DE F G HI J K LM N O P假设有一个图片，4*4，共16个像素，分别用AL来代表。我们先观察这个图片，只有中间的F,G,J,K这四个像素的“邻居”是全的。为了简便起见，我们只处理这4个像素，因为在实际的图片中由于图片的大小都很多像素组成，所以周围的一圈像素不做处理不会影响到最终的效果。先计算差值: Delta= F - (A+B+C+E+G+I+J+K) / 8(A+B+C+E+G+I+J+K) / 8就是F周围的像素的平均值，将这个平均值乘以一个系数再加到F上，就得到了一个新的F值：F=

16、F + Delta * Alpha这个系数Alpha就是锐化度，改变这个系数就能得到不同的锐化效果。不过一般都是取得比较小的，如：0.3于是，我们只要使用两个循环来遍历整个图片的像素值（去除边界）就能得到一个锐化的效果了。但是大家或许会发现在处理后面几个点的时候，前面的点的值已经不是原来的值了，比如处理G的时候，需要用到F的值，而F则已经被改变，并且F的改变又和G的值有关系，这样就会变成一种循环引用。为了避免整个问题，这里给出一个改良的方法：A B C DE F G HI J K LM N O P我们从A点开始做，将差值计算方法改成: Delta= A - (B+E+F) / 3F=F + D

17、elta * Alpha按照从左到右，从上到下的顺序来扫描所有像素，这时在计算中就不会遇到已经被处理过的像素了，并且因为减少了参与运算的像素，整个处理过程也得以加快。按照我们在VB图像处理之像素的获取和输出中已经得到的像素数组。我们可以这样写：Public Sub Sharp(Optional ByVal SharpDgree As Single = 0.3)Dim X As LongDim Y As LongDim Ix As LongDim Iy As LongDim Diff As LongDim Diff1 As LongDim Div1 As SingleDim Div2 As Si

18、ngleDim Max As LongOn Error GoTo ErrLineMax = 255Done = FalseTimeFilter = timeGetTimeTemplateSize = 1Sensitivity = Sensitivity * 9Div1 = 1 + SharpDgreeDiv2 = -SharpDgree / 3For X = 0 To OutPutWid - 1 For Y = 0 To OutPutHei -1RR = ColOut(0, X, Y) * Div1GG = ColOut(1, X, Y) * Div1BB = ColOut(2, X, Y)

19、* Div1Ix = X + 1 Iy = Y + 1 R = ColOut(0, Ix, Iy)R = R + ColOut(0, X, Iy) + ColOut(0, Ix, Y)G = ColOut(1, Ix, Iy)G = G + ColOut(1, X, Iy) + ColOut(1, Ix, Y)B = ColOut(2, Ix, Iy)B = B + ColOut(2, X, Iy) + ColOut(2, Ix, Y)R = R * Div2G = G * Div2B = B * Div2RR = RR + RGG = GG + GBB = BB + BIf RR Max T

20、hen RR = MaxIf GG Max Then GG = MaxIf BB Max Then BB = MaxColOut(0, X, Y) = RRColOut(1, X, Y) = GGColOut(2, X, Y) = BBNextNextDone = TrueTimeFilter = timeGetTime - TimeFilterExit SubErrLine:Done = TrueMsgBox Err.DescriptionEnd Sub因为在计算新的像素的过程中会出现新的值大于255或小于0的情况，因此必须在计算完成后判断。所用到的全局变量：Public TimeFilte

21、r As Long 用于记录滤镜处理所花费的时间 Dim RR As Long 用于保存红色分量Dim GG As Long 用于保存绿色分量Dim BB As Long 用于保存蓝色分量原图：锐化效果：1 2 下一页上次讲到了用DIB方法来获取图像的像素。从这次开始将如果运用已经得到的像素来处理图像。图像插值放大的方法有很多，最主要的有二次线性插值和三次线性插值这两种。这次我把自己的程序中所用的二次线性插值的算法公布给大家，希望对各位要使用VB写类似程序的朋友有所帮助。程序中用到的API、数据类型、全局变量的定义请参考上一篇：VB实现图像在数据库的存储与显示 Public Sub ZoomI

22、mage(ByVal OutPutWidth As Long, ByVal OutputHeight As Long)Dim I As LongDim L As LongDim X As LongDim Y As LongDim Xb As LongDim Yb As LongDim Xe As LongDim Ye As LongDim M As IntegerDim N As IntegerDim CurR As LongDim CurG As LongDim CurB As LongDim NxtR As IntegerDim NxtG As IntegerDim NxtB As Int

23、egerDim DR As SingleDim DG As SingleDim DB As SingleDim DRt As SingleDim DGt As SingleDim DBt As SingleDim Xratio As SingleDim Yratio As SingleDim CurStep As SingleDim NxtStep As SingleDim NegN As SingleOn Error GoTo ErrLineIf Not CanZoom Then Exit SubDone = FalseOutPutWid = OutPutWidth - 1OutPutHei

24、 = OutputHeight - 1I = (Bits 8) - 1ReDim ColTmp(I, InPutWid, OutPutHei) 先从Y方向进行缩放处理，结果保存在此中间数组内ReDim ColOut(I, OutPutWid, OutPutHei)Xratio = OutPutWid / InPutWidYratio = OutPutHei / InPutHeiTimeZoom = timeGetTimeNegN = 1 / Int(Yratio + 1)For X = 0 To InPutWidCurR = ColVal(0, X, 0)CurG = ColVal(1, X,

25、 0)CurB = ColVal(2, X, 0)CurStep = 0NxtStep = 0For Y = 0 To InPutHei - 1NxtStep = CurStep + YratioYb = CurStepYe = NxtStepN = Ye - YbColTmp(0, X, Yb) = CurRColTmp(1, X, Yb) = CurGColTmp(2, X, Yb) = CurBM = Y + 1NxtR = ColVal(0, X, M)NxtG = ColVal(1, X, M)NxtB = ColVal(2, X, M)If N 1 ThenDRt = (NxtR

26、- CurR) * NegNDGt = (NxtG - CurG) * NegNDBt = (NxtB - CurB) * NegNDR = 0DG = 0DB = 0For L = Yb + 1 To Ye - 1DR = DR + DRtDG = DG + DGtDB = DB + DBtColTmp(0, X, L) = CurR + DR ColTmp(1, X, L) = CurG + DG ColTmp(2, X, L) = CurB + DB NextEnd IfCurStep = NxtStepCurR = NxtRCurG = NxtGCurB = NxtBNextColTm

27、p(0, X, OutPutHei) = NxtRColTmp(1, X, OutPutHei) = NxtGColTmp(2, X, OutPutHei) = NxtBNextNegN = 1 / Int(Xratio + 1)For Y = 0 To OutPutHeiCurR = ColTmp(0, 0, Y)CurG = ColTmp(1, 0, Y)CurB = ColTmp(2, 0, Y)CurStep = 0NxtStep = 0For X = 0 To InPutWid - 1NxtStep = CurStep + XratioXb = CurStepXe = NxtStep

28、N = Xe - XbColOut(0, Xb, Y) = CurRColOut(1, Xb, Y) = CurGColOut(2, Xb, Y) = CurBM = X + 1NxtR = ColTmp(0, M, Y)NxtG = ColTmp(1, M, Y)NxtB = ColTmp(2, M, Y)If N 1 ThenDRt = (NxtR - CurR) * NegNDGt = (NxtG - CurG) * NegNDBt = (NxtB - CurB) * NegNDR = 0DG = 0DB = 0For L = Xb + 1 To Xe - 1DR = DR + DRtDG = DG + DGtDB = DB + DBtColOut(