java-face人脸识别程序（代码）

import java.applet.*; import java.awt.*; import java.awt.image.*; import java.net.*; import java.util.*; import java.io.*; /** *ImageXor is an algorithm to apply logical XOR/XNOR between two images *@author:Judy Robertson, SELLIC OnLine *@author:Neil Brown, DAI *@author:Timothy Sharman *@see code.iface.imagexor */ public class ImageXor extends Thread{ //the width of the input images in pixels private int i1_w; private int i2_w; //the width and height of the output image private int d_w; private int d_h; private int[] dest_1d; private boolean is_colored; /** *Constructs a new Image Xor *@param firstwidth The width of the iorst input image *@param secondwidth The width of the second input image */ public ImageXor(int firstwidth, int secondwidth){ i1_w = firstwidth; i2_w = secondwidth; } /** *Applies the image XOR operator on the specified image arrays, with the specified offset and scale value *@param src1_1d The first source image as a pixel array *@param src2_1d The second source image as a pixel array *@param XNOR Boolean to indicate wether we should XNOR rather than XOR *@param width width of the destination image in pixels *@param height height of the destination image in pixels *@param oset The offset value *@param scale The scale value *@return A pixel array containing the X(N)OR of the two input images */ //Bob's image xor algorithm.. /*a) assume the image is grey level (hence RR=GG=BB) b) use value &0x000000ff to get the BB value c) do this for both input images d) apply the operation (eg XOR) e) add oset and then scale f) clip to lie from 0 to 255. Call this value 0xCD g) create int value 0xffCDCDCD */ public int [] doXor(int [] src1_1d, int [] src2_1d, boolean XNOR, int width, int height, float oset, float scale){ int place1 = -1; int place2 = -1; int src1rgb = 0; int src2rgb = 0; int result = 0; //Get size of image and make 1d_arrays d_w = width; d_h = height; dest_1d = new int[d_w*d_h]; boolean firstwider = false; boolean secondwider = false; int wrap; if (i1_w > d_w){ wrap = ((i1_w + 1) - d_w); firstwider = true; } else if (i2_w > d_w){ wrap = ((i2_w + 1) - d_w); secondwider = true; } else { wrap = 0; } //if you know there is no wrap around, you can save yourself some time if (wrap == 0) { for (int i=0; i< dest_1d. length ; i++){ src2rgb = src2_1d[i] & 0x000000ff; src1rgb = src1_1d[i] & 0x000000ff; /* Due to Java not having unsigned ints, the ~ will always give a negative result. As we a dealing with only the bottom 8 bits of the int, I will just mask out the bottom bits. Neil. */ if (XNOR) { result = (int) ((scale * (float)(~(src1rgb ^ src2rgb) & 0xff))+oset); } else { result = (int) ((scale * (float) ( src1rgb ^ src2rgb )) + oset); } //clip to 0 ... 256 if (result < 0){ result = 0; } else if (result > 255){ result = 255; } //create an int value for dest_1d dest_1d[i ] = 0xff000000 | (result + (result << 16) + (result << 8)); } return dest_1d; } else { for (int i=0; i< dest_1d. length ; i++){ //we might need to skip out some pixels which aren't in the overlap area if ((i %d_w ) == 0 ) { if ( i == 0 ){ place1 = 0; place2 = 0; } else if (secondwider) { place2 = place2 + wrap; place1 ++; } else { place1 = place1 + wrap; place2 ++; } } else{ place2 ++; place1 ++; } src2rgb = src2_1d[place2] & 0x000000ff; src1rgb = src1_1d[place1] & 0x000000ff; /* Due to Java not having unsigned ints, the ~ will always give a negative result. As we a dealing with only the bottom 8 bits of the int, I will just mask out the bottom bits. Neil. */ if (XNOR) { result = (int) ((scale * (float) (~(src1rgb ^ src2rgb) & 0xff))+ oset); } else { result = (int) ((scale * (float) ( src1rgb ^ src2rgb )) + oset); } //clip to 0 ... 256 if (result < 0){ result = 0; } else if (result > 255){ result = 255; } //create an int value for dest_1d dest_1d[i ] = 0xff000000 | (result + (result << 16) + (result << 8)); } return dest_1d; } } /** *XORs the specified input image by the constant, also adds offset * *@param src1_1d The input pixel array *@param constant The constant value to AND every pixel in the input array by *@param XNOR Boolean to indicate wether we should XNOR rather than XOR *@param width width of the destination image in pixels *@param height height of the destination image in pixels *@param oset The offset value *@param scale The scale value *@return A pixel array with the constant value X(N)ORed with every pixel in the input array */ public int [] doXor(int [] src1_1d, int constant, boolean XNOR, int width, int height, float oset, float scale ){ //Get size of image and make 1d_arrays d_w = width; d_h = height; dest_1d = new int[d_w*d_h]; int src1rgb; int result = 0; /* Due to Java not having unsigned ints, the ~ will always give a negative result. As we a dealing with only the bottom 8 bits of the int, I will just mask out the bottom bits. Neil. */ //now do the XOR on the input image for (int i=0; i< src1_1d. length; i++){ src1rgb = src1_1d[i] & 0x000000ff; if(XNOR) { result = (int) (( scale * ((float) (~(src1rgb ^ constant) &0xff))) + oset); } else { result = (int) (( scale * ((float) ( src1rgb ^ constant ))) + oset); } if (result > 255){ result = 255; } if (result < 0){ result = 0; } dest_1d[i] = 0xff000000 | (result + (result << 16) + (result << 8)); } return dest_1d; } }