C++ 模板复杂图像读取类执行时间慢，声明和实现分离

我正在尝试制作读取复杂图像的二进制文件。我使用VisualStudio 2010。这是代码1：

      #include <stdio.h>
      #include <conio.h>
      #include <iostream>
      #include <fstream>
      #include <math.h>
      #include <stdlib.h>
      #define NX 21843
      #define NY 22380
      #include <windows.h>
    class CTimer
    {
    public:
        CTimer() 
        {
            QueryPerformanceFrequency(&mqFreq);
        }
        ~CTimer() {}
        void Start() 
        {
            QueryPerformanceCounter(&mqStart);
        }
        void End() 
        {
            QueryPerformanceCounter(&mqEnd);
        }
        double GetTimeInSeconds()
        {
            return (mqEnd.QuadPart - mqStart.QuadPart)/static_cast<double>(mqFreq.QuadPart);
        }
    private:
        LARGE_INTEGER mqStart;
        LARGE_INTEGER mqEnd;
        LARGE_INTEGER mqFreq;
    };
    using namespace std;
int main(void)
{
    CTimer Time;
    Time.Start();
    ifstream INFILE;
    INFILE.open("e:\WORK\CIMG.bindat", ios::binary);
    short ***im_ptr;
    im_ptr = new short** [2];
    for (int i=0; i<2; i++){
        im_ptr[i] = new short* [NX];
        for (int j=0; j<NX; j++){
            im_ptr[i][j] = new short [NY];
        }
    }
    for(int i=0; i<2; i++){
        for (int j=0; j<NX; j++){
            for (int k=0; k<NY; k++){
                int index = i*NY*NX + j*NY + k;
                INFILE.read((char*)&im_ptr[i][j][k], sizeof(short));
                if (index == 2*NX*NY-1){
                    INFILE.close();}
            }
        }
    } 
    for (int i=0; i<2; i++){
        for (int j=0; j<NX; j++){
            delete[] im_ptr[i][j];
        }
        delete[] im_ptr[i];
    }
    delete[] im_ptr;
    Time.End();
    std::cout << Time.GetTimeInSeconds() << " seconds" << std::endl;
    std::cout << "nPress any key to exit" << std::endl;
    getch();
    return 0;
}

此代码工作正常，但仅适用于 SHORT 类型的图像。

我决定做模板类，它允许读取许多不同数据类型的图像，并将声明和实现分开。这是代码2：

stdafx.h 的 - 声明我的班级

#pragma once
#include "targetver.h"
#include <stdio.h>
#include <tchar.h>
template<typename DataType> 
class SARDataLoader
{
private:
    int day,month,year; 
public:
    DataType*** LoadComplexImage(int, int);
    int DeleteImageFromMemory(DataType***, int);
}; 

SARDataLoader_C.cpp - 实现我的类

#include "stdafx.h"
#include <iostream>
#include <fstream>
#include <stdlib.h>
using namespace std;
template<typename DataType> 
DataType*** SARDataLoader<DataType>::LoadComplexImage(int NX, int NY){
    using namespace std;
    ifstream INFILE;
    INFILE.open(""e:\WORK\CIMG.bindat"", ios::binary);
    DataType ***im_ptr;
    im_ptr = new DataType** [2];
    for (int i=0; i<2; i++){
        im_ptr[i] = new DataType* [NX];
        for (int j=0; j<NX; j++){
            im_ptr[i][j] = new DataType [NY];
        }
    }
    for(int i=0; i<2; i++){
        for (int j=0; j<NX; j++){
            for (int k=0; k<NY; k++){
                int index = i*NY*NX + j*NY + k;
                INFILE.read((char*)&im_ptr[i][j][k], sizeof(DataType));
                if (index == 2*NX*NY-1){
                    INFILE.close();}
            }
        }
    }
    return im_ptr;
};
template<typename DataType> int SARDataLoader<DataType>::DeleteImageFromMemory(DataType*** im_ptr, int NX){
    for (int i=0; i<2; i++){
        for (int j=0; j<NX; j++){
            delete[] im_ptr[i][j];
        }
        delete[] im_ptr[i];
    }
    delete[] im_ptr;
    return 112345;
};
template class SARDataLoader<short>;//explicit instantiation short

主.cpp - 使用我的类

#include "stdafx.h"
#include <stdio.h>
#include <conio.h>
#include <iostream>
#include <fstream>
#include <math.h>
#define NX 21843
#define NY 22380
#include <windows.h>
class CTimer
{
public:
    CTimer() 
    {
        QueryPerformanceFrequency(&mqFreq);
    }
    ~CTimer() {}
    void Start() 
    {
        QueryPerformanceCounter(&mqStart);
    }
    void End() 
    {
        QueryPerformanceCounter(&mqEnd);
    }
    double GetTimeInSeconds()
    {
        return (mqEnd.QuadPart - mqStart.QuadPart)/static_cast<double>(mqFreq.QuadPart);
    }
private:
    LARGE_INTEGER mqStart;
    LARGE_INTEGER mqEnd;
    LARGE_INTEGER mqFreq;
};
using namespace std;
int main(void)
{
    CTimer Time;
    Time.Start();
    std::cout <<"nIMAGE1 Loading..."<< std::endl;
    SARDataLoader<short> IMG_LOADER;
    short ***im_ptr = IMG_LOADER.LoadComplexImage(NX, NY);
    int CODE = IMG_LOADER.DeleteImageFromMemory(im_ptr, NX);
    Time.End();
    double run_time = Time.GetTimeInSeconds();
    std::cout <<CODE<<"nRUNTIME "<<run_time<<" seconds"<<"n...Press any key to exit...n"<<std::endl;
    getch();
    return 0;
}

此代码有效，并且执行与 code1 相同的操作。我使用 CTimer 类进行代码 1 和代码 2 执行时间测量。它说 code2 的执行时间几乎是 code1 的三倍。谁能告诉我是什么原因造成的以及如何解决这个问题？