直接从文件中读取大量bmp数据

Question

I am working on an application that has to deal with large bmp files that, quite often, are too large to fit into memory. As part of the software we are developing at project creation my application converts the data from a bmp file to a different format that allows easy retrieval in sections.

Currently, since the file is often too large to fit in memory, the software reads a section of byte data directly from the file, processes it, and moves onto the next section. The code for reading the file is similar to that shown below (simplified for clarity):

FileStream fs = File.OpenRead(fileName);
fs.Seek(sectionStart, SeekOrigin.Begin); 

currentSectionAsBytes = new byte[sectionSize];
fs.Read(currentSectionAsBytes, 0, currentSectionAsBytes.Length);

This has worked fine up until now since the files we work with all have a width divisible by 4, and therefore the bmp files have no padding.

Lately, we've been working with different sized images which have padding at the end of each row of data, causing the acquired data to be erroneous.

My question is, does anyone know of any better ways of acquiring bmp data directly from file. As previously stated I can't read the Bitmap from the file due to the large file size. Currently my best idea is, after getting the currentSectionAsBytes, to calculate the padding size and manually remove it. This just seems too cumbersome and complicated. There must be a better solution.

Answer 1

"Manually removing" something from an array is a bad idea, when you are facing the lack of memory (since it will lead to creating another array). I think, the right approach to your issue is to add two extra parameters to your processing method, so its siganture will look like:

void ProcessSection(byte[] data, int offset, int count);

And then iterate data not from 0 to data.Length , but from offset to count . This is a common way to deal with arrays in .Net (look at fs.Read() method for example), there is not much else that can be done. And yes, you will have to calculate offset/count according to your bitmap properties.

Ofc, alternatively, you can account for an offset, when you are calling fs.Seek method (eg when you calculate sectionStart ), but i think the first apporach is more universal.