hauke96 / image2tiles Goto Github PK

Currently the code is not very testable and has to be split up into functions that can be tested separately. Think of tests using e.g. cunit as framework.

Currently the flags -h and --help are unknown. Change that. Currently I don't know a fancy way to add a help message (with correct formatting and stuff), maybe there is a good way?

The tiles are in the size used in the original image. This may be 134 pixel but normally a tile as the size 256. Scale each tile to the according size.

There's already a detailed documentation for the calculation of the next higher tile (e.g. from zoom-level 13 to zoom-level 12). Add additional ones for other non-trivial calculations.

Instead of specifying the first tile position inside the image with some sort of index, it would be nice to just specify die lat-long position in combination with the maximum zoon level. A calculation of the first tile position and X/Y/Z coordinates can be done in the code (as seen here).

Zoom levels are currently not used. An additional for loop is needed to loop over the zoom levels. Also the tile size (in the image and from the output file), X and Y coordinate as well as the position in the image have to be calculated based on the zoom level.

Refactor the usage of parameters in two steps:

• Use variables instead of hard-coded values
• Use arguments from the CLI

Handle errors (e.g. missing parameter or wrong format) when parsing the CLI arguments.

Currently the code base is not that large. Nevertheless I don't have strong arguments against splitting the code into several files. At least a debug.h for several makros would be nice. Maybe separate files for I/O, calculations and parsing might be nice.

When #1 is done, create a sample that cuts the sample/sample.jpg image into the according tiles.

The README should contain:

• General description
• Usage information
• Requirenments (e.g. opencv and its possibly wrong location)
• Build instructions
• ...?