http://www.ruanyifeng.com/blog/2013/03/similar_image_search_part_ii.html c++的实现

similar Bmp.cpp

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <iostream>
using namespace std;
typedef struct
{
    char type1;
    char type2;
}BmpFileHead;
typedef struct
{
    unsigned int imageSize;
    unsigned int blank;
    unsigned int startPosition;
    unsigned int    length;
    unsigned int    width;
    unsigned int    height;
    unsigned short    colorPlane;
    unsigned short    bitColor;
    unsigned int    zipFormat;
    unsigned int    realSize;
    unsigned int    xPels;
    unsigned int    yPels;
    unsigned int    colorUse;
    unsigned int    colorImportant;
}BmpInfoHead;
typedef struct{
    unsigned char G;
    unsigned char B;
    unsigned char R;
    unsigned char A;
}RGBA;

typedef struct
{
    BmpFileHead fileHead;
    BmpInfoHead infoHead;
    unsigned char * pData;
}BmpFile;

BmpFile* LoadBMPImage( const char* filename)
{
    int channels;
    FILE * p;
    BmpFile *pBmpFile = NULL;
    int step, modbytes, idx;

    if( !filename || strlen(filename) == 0 )
    {
        fprintf(stderr, "null filename\n");
        return NULL;
    }
    //打开文件
    p = fopen(filename, "rb");
    if(!p)
    {
        fprintf(stderr, "Can not open file %s\n", filename);
        return NULL;
    }
    
    pBmpFile = (BmpFile*)malloc( sizeof(BmpFile));
    
    //fread(&bmpfilehead, sizeof(BmpFileHead), 1, p);
    //从文件中读BmpFileHead，在一些系统中sizeof(BmpFileHead)=4
    //所以不能使用上面那行代码，而要使用下面这行
    fread(&(pBmpFile->fileHead), 2, 1, p);
    
    //从文件中读取BmpInfoHead
    fread(&(pBmpFile->infoHead), sizeof(BmpInfoHead), 1, p);
    
    //如果文件的前两个字节不是'BM'，则不是BMP文件
    if( (pBmpFile->fileHead.type1 != 'B' && pBmpFile->fileHead.type1 != 'b') ||
        (pBmpFile->fileHead.type2 != 'M' && pBmpFile->fileHead.type2 != 'm') )
    {
        fprintf(stderr, "The file is not a BMP file" );
        fclose(p);
        free(pBmpFile);
        return NULL;
    }
    
    //对读入的数据进行检查
    //对于Gray图像, bitColor=8;
    //对于RGB图像, bitColor=24
    channels = pBmpFile->infoHead.bitColor / 8;
    if( channels != 1 && channels != 3)
    {   //只支持1个或3个通道的图像
        fprintf(stderr, "Only 1 or 3 channels images can be loaded." );
        fclose(p);
        free(pBmpFile);
        return NULL;
    }

    //检查宽度和高度是否合理
    if(pBmpFile->infoHead.width <= 0 || pBmpFile->infoHead.height <=0)
    {
        fprintf(stderr, "Bad image size");
        fclose(p);
        free(pBmpFile);
        return NULL;
    }
    
    //每行像素占的字节数为4的倍数，计算出这个数值step
    modbytes = (pBmpFile->infoHead.width * channels)%4;
    step = modbytes ? ((pBmpFile->infoHead.width*channels)-modbytes+4)  : (pBmpFile->infoHead.width*channels);

    pBmpFile->pData = (unsigned char*)malloc(step * pBmpFile->infoHead.height);

    //跳到文件中的数据区开始位置
    fseek(p, pBmpFile->infoHead.startPosition, SEEK_SET);
    
    //将所有图像数据读入，由于BMP以底端为第一行存储，
    //而图像一般以上端为第一行存储
    for(idx=pBmpFile->infoHead.height-1; idx >= 0 ; idx--)
    {
        int n = fread(pBmpFile->pData + step*idx, step, 1, p);
        if(n!=1)
        {
            fprintf(stderr, "read image data error.\n");
            break;
        }
    }
    
    fclose(p);

    return pBmpFile;
}


int SaveBMPImage( const char* filename, const BmpFile * pBmpFile )
{
    FILE * p;
    int channels;
    int modbytes, step;
   
    if( !filename || strlen(filename) == 0 )
    {
        fprintf(stderr, "null filename\n");
        return -1;
    }

    if(!pBmpFile)
    {
        fprintf(stderr, "null BMP file data\n");
        return -1;
    }
    
   //读取通道数目
    channels = pBmpFile->infoHead.bitColor / 8;
    //每行像素占的字节数为4的倍数，计算出这个数值step
    modbytes = (pBmpFile->infoHead.width * channels)%4;
    step = modbytes ? ((pBmpFile->infoHead.width*channels)-modbytes+4)  : (pBmpFile->infoHead.width*channels);
    
    p = fopen(filename, "wb");
    if(!p)
    {
        fprintf(stderr, "Can not create file" );
        return -1;
    }
    
    //写入数据头信息
    fwrite(&(pBmpFile->fileHead), sizeof(BmpFileHead), 1, p);
    fwrite(&(pBmpFile->infoHead), sizeof(BmpInfoHead), 1, p);

    
    //写入数据信息
    if(channels==1)
    {
        RGBA rgba;
        for(int idx =0; idx<256; idx++)
        {
            rgba.R=idx;
            rgba.G=idx;
            rgba.B=idx;
            rgba.A = 0;
            fwrite(&rgba, sizeof(RGBA), 1, p);
        }
    }
    //写入数据
    for(int idx = pBmpFile->infoHead.height-1; idx >=0 ; idx--)
        fwrite(pBmpFile->pData + step*idx, step, 1, p);
    
    fclose(p);
    
    return 0;
}

//GetStep 函数返回 BmpFile->pData 表示的矩阵的宽度
int GetStep(BmpFile* file, int channel){
    int modbytes = (file->infoHead.width * channel) %4 ;
    int step = modbytes ? ((file->infoHead.width * channel) - modbytes + 4) : (file->infoHead.width * channel);
    return step;
}

// 去色并转变为灰度图片
void RemoveColor(BmpFile* file){
    int channels = file->infoHead.bitColor / 8;
    
    if(channels != 3){
        cout << "Bmp file's channel must be 3" << endl;
        return ;
    }

    int i, j;
    int height = file->infoHead.height;
    int step = GetStep(file, 3);
    unsigned char R, G, B;
    double Gray;
    unsigned char* newData = new unsigned char[height * step / 3];
    int index=0;
    for(i=0;i<height;i++){
        for(j=0;j<step;j+=3){
            R = file->pData[i*step + j + 2];
            G = file->pData[i*step + j + 1];
            B = file->pData[i*step + j];
            Gray = 0.299 * R + 0.587 * G + 0.114 * B;
            newData[index++] = (unsigned char)Gray;
        }
    }
    file->infoHead.bitColor = 8;
    file->pData = newData;
}

void binary_Bmp(BmpFile* file){
    int n = file->infoHead.height * GetStep(file, 1);    // 总数
    int n1, n2;         // 大于阈值的像素点数量、小于阈的像素点数量
    int s1, s2;         // 大于阈值的像素总和、小于阈值的像素总和
    double w1, w2;         // 权值
    double u1, u2;         // 平均值

    double maxBetween = 0.0;
    double between;
    int threshold;

    for(int t=0; t<256; ++t)
    {
        n1 = 0;
		n2 = 0;
		s1 = 0;
		s2 = 0;
        for(int i=0; i<n; ++i)
        {
            if((int)file->pData[i] < t)
            {
                ++n1;
                s1 = s1 + (int)file->pData[i];
            }
            else
            {
                ++n2;
                s2 = s2 + (int)file->pData[i];
            }
        }

        w1 = (double)n1 / (double)n;
        w2 = (double)n2 / (double)n;
		u1 = (double)s1 / (double)n1;
        u2 = (double)s2 / (double)n2;

        between =  w1 * w2 * (u1 - u2) * (u1 - u2);
        
        if( n1 && n2 && maxBetween < between)
        {
            maxBetween = between;
            threshold = t;
        }  
       
    }
    for(int i=0; i<n; ++i)
    {
		file->pData[i] = file->pData[i] > threshold ? (char)255 : (char)0;
    }
}

int compare(BmpFile* one, BmpFile* two)
{
	int n1 = one->infoHead.height * GetStep(one, 1);
	int n2 = two->infoHead.height * GetStep(two, 1);
	if(n1 != n2)
	{
		return -1;
		//error
	}
	int r=0;
	for(int i=0; i<n1; ++i)
	{
		r += (int)one->pData[i] ^ (int)two->pData[i];
	}
	return r;
	
}

int main()
{
    BmpFile* origin_bmp = LoadBMPImage("origin-rgb.bmp");
	BmpFile* com_bmp_1 = LoadBMPImage("origin-rgb-1.bmp");
	BmpFile* com_bmp_2 = LoadBMPImage("origin-rgb-2.bmp");

    RemoveColor(origin_bmp);
	binary_Bmp(origin_bmp);

	RemoveColor(com_bmp_1);
	binary_Bmp(com_bmp_1);

	RemoveColor(com_bmp_2);
	binary_Bmp(com_bmp_2);
	
	int c1 = compare(origin_bmp, com_bmp_1);
	int c2 = compare(origin_bmp, com_bmp_2);

	if(c1 < c2){
		cout << "similar: Pic1" << endl;
	}
	if(c1 == c2){
		cout << "Equal" << endl;
	}
	if(c1 > c2){
		cout << "similar: Pic2" << endl;
	}
	cout << "c1=" << c1 << " c2=" << c2 << endl;

	//SaveBMPImage("D:\\Projects\\cpp\\test\\Debug\\to_bin_bmp.bmp", origin_bmp);


    return 1;
}