Code Works on Mac but not on Windows C++- C6001 Using Uninitialized memory Error

Its basically a fairly simple code in c++ Which i working perfectly fine on MAC OS but on windows it's giving hard time as I not well aware how does windows Api is treating the code.

Code as follows (Where problem Starts) :

unsigned char*** output_data = new unsigned char** [8300];

for (int32_t i = 0; i < 8300; i++) {
    output_data[i] = new unsigned char* [192];
    for (int32_t j = 0; j < 192; j++) {
        output_data[i][j] = new unsigned char[4];
    }
}

//for each strip, except the last
for (int16_t s = 0; s < (number_strips); ++s) {

    //clean the output array
    for (int16_t i = 0; i < 8300; ++i) {
        for (int16_t j = 0; j < 192; ++j) {
            for (int16_t k = 0; k < 4; ++k) {
                output_data[i][j][k] = 255;
            }
        }
    }

    for (int16_t j = 0; j < 192; ++j) {

        //read in the new data for 192 rows
        if (s < (number_strips - 1)) fread(input_data, 1, row_padded, iFile);

        //This is what to do on the last strip, which may be unfilled
        if (s == (number_strips - 1)) {
            if ((j) < extra_rows) {
                fread(input_data, 1, row_padded, iFile);
            }
            else {
                for (uint32_t i = 0; i < row_padded; ++i) {
                    input_data[i] = static_cast<unsigned char>(whitePixel);
                }
            }
        }

        //rewrite the data to the output array, flipping the order
        //note we use width instead of row_padded here to remove the row_padded bytes
        for (int16_t i = kOffset; i < out_array_size; ++i) {
            output_data[i][j][0] = input_data[(3 * (i - kOffset))];
            output_data[i][j][1] = input_data[(3 * (i - kOffset) + 1)];
            output_data[i][j][2] = input_data[(3 * (i - kOffset) + 2)];
        }

        //apply the CYMK mask
        for (int16_t i = 0; i < (out_array_size + kOffset); ++i) {

            holderR = static_cast<float>(output_data[i][j][0]);
            holderG = static_cast<float>(output_data[i][j][1]);
            holderB = static_cast<float>(output_data[i][j][2]);

            holderR = (holderR / 255);
            holderG = (holderG / 255);
            holderB = (holderB / 255);

            holderK = holderR;
            if (holderK < holderG) holderK = holderG;
            if (holderK < holderB) holderK = holderB;
            holderK = 1 - holderK;

            holderC = 255 * (1 - holderR - holderK) / (1 - holderK);
            holderY = 255 * (1 - holderG - holderK) / (1 - holderK);
            holderM = 255 * (1 - holderB - holderK) / (1 - holderK);
            holderK = 255 * holderK;

            output_data[i][j][0] = static_cast<unsigned char>(holderM);
            output_data[i][j][1] = static_cast<unsigned char>(holderC);
            output_data[i][j][2] = static_cast<unsigned char>(holderY);
            output_data[i][j][3] = static_cast<unsigned char>(holderK);
        }

        //transpose data in the output array based on the offsets
        for (int16_t i = 0; i < (out_array_size); i++) {
            output_data[i][j][0] = output_data[i + kOffset][j][0];
            output_data[i][j][1] = output_data[i + mOffset][j][1];
            output_data[i][j][2] = output_data[i + yOffset][j][2];
            output_data[i][j][3] = output_data[i + cOffset][j][3];
        }

    } //finish of function for each row

Compiler ignores this loop and gives following error :

Warning C6001   Using uninitialized memory '*output_data[i]'.   
*Output_data[i] not initialized 
*skip this loop assume s<(number_strips is false)

I have absolutely no idea how it behaves like this while in Mac it works without any hassle. Any help would be much help full :)