Nerfplusplus parameter format sorting

​​​​​​GitHub - Kai-46/nerfplusplus: improves over nerf in 360 capture of unbounded scenes

Output Format — COLMAP 3.7 documentation

We're running colmap It's the back , Three parameters are output ：camera,images,points3D

  In the official interpretation , The structure of these three parameters is as follows ：

         And in the nerfplusplus Format data set , What we use is such a data set ：

        

  see pose file , The discovery parameter is composed of the following format ：

 （ To make it easier to watch, I hit enter ）

  see intrinsics file , It is found that the parameters are composed of the following formats ：

  Now I share my data conversion methods , The specific steps are: ： obtain colmap file — Import ecxel— Invalid information filtering — Matrix transformation — Matrix adjustment — Segmented export , The whole process is as follows ：

First of all pose file

（1）colmap export ： Select export as txt Format

（2） Import excel, Invalid information filtering ： Select from text in the data , This can be done in excel Auto wrap in . Import excel And then we found out , The information of each image is composed of two lines , The first line is the image pose information we need , The other line is the pixel information of the image itself , Let's get rid of the useless second line , Leave only useful information . Do this for every picture .

（3） Matrix transformation ： This is the most critical ,colmap In the output mode of , Posture is quaternion QW,QX,QY,QZ Composed of . They and 3*3 The matrix transformation relation of is as follows ：

        (27 Bar message ) From rotation matrix to quaternion _loongkingwhat The blog of -CSDN Blog _ Rotation matrix to quaternion

          Use excel It can be easily transformed , after , We readjust the table according to the following format ：

（4） Export to TXT Format （ I wrote a simple code ）：

import xlrd

def read_excel_data(filepath):
    workbook = xlrd.open_workbook(filepath)
    sheet = workbook.sheet_by_index(0)
    i = '1';
    for index in range(1, sheet.nrows):
        row_value = sheet.row_values(index)
        print(row_value)
    
        file_handle=open('%s.txt'%(i),mode='w')
        file_handle.seek(0,0)

        for j in range(16):
            file_handle.write(str(row_value[j]))
            file_handle.write(' ')
        j=0
        
        file_handle.close()
        i = str(int(i) + 1)
        
if __name__ == '__main__':
    read_excel_data('round8.xls')