JPEG Compression Algorithm Comparison Testing

Purpose

The purpose of this experiment is to choose a compression algorithm for MOCI Sat's images to be transmitted to the ground station. In addition to the compression algorithm itself, we will also need to choose the best starting file type and ending file type that will result in the best quality of image balanced against the size of the file and the time required to convert between the file types.

Overview

This program will gather data regarding the size of compressed images generated by different compression algorithms and the time taken to do so. The purpose of this is to choose a compression algorithm to use on the satellite images taken by MOCI sat. This bash script will take in many images of varying file sizes and containing varying geographic features (such as images of the ocean, coastline, and land, including forests, planes, and cities). The compression algorithm will then be run on each image, and the resulting file size and runtime will be written to the data csv. This csv will contain all the file sizes and runtimes for each file, each compression algorithm, and each start and end file type, which can then be graphed and compared.

Compression Algorithms Tested

Test Run 1 -- June 2016

The first two algorithms compared with be two versions of OpenJPEG. OpenJPEG, version 1 -- ocomp -- https://github.com/uclouvain/openjpeg/ OpenJPEG, version 2 -- pcomp -- https://github.com/planetlabs/openjpeg/

File Types Used

Test Run 1 -- June 2016

Start Types

We will start with the following types, which will be converted into each end type.

• .png
• .tif
• .bmp
• .raw

End Types

We will convert each starting type, using each algorithm, into the following types.

• .j2k
• .j2c
• .jp2

Input Test Files

The input will start out as 30 test files containing a variety of geographic features, including:

• Ocean
• Coast and ocean
• Fall foliage
• Spring foliage
• City All files will start out as .raw files and will then be converted to each of the other input file types. They will then be resized so that there are a variety of file sizes for each type, containing each geographic feature. Input test files can be found in the following path: ./img/in/it-#/[png/tif/bmp/raw]/ Each iteration will have file descriptions in the file info.csv contained in the path ./img/in/it-#/

Compressed Image Files

Each image file will be compressed to each of the end file types using both compression algorithms. This means that for every file in the input set, there will be six files in the output set. These files will be contained in the following path: ./img/out/it-#/

Test Out CSV

The bash script will create a .csv file that will contain the starting file's information, especially the file size and a description of the image. The output CSV can be found at the following path: ./img/out/it-#/out.csv

Usage

To run the test script on a directory of images using a compression algorithm and outputting to a directory, use the command: ./test_comp.sh # <start_type> <PATH to compression algorithm> [end_types] Where:

• # is the iteration number
• <start_type> is the starting file type
• <PATH to compression algorithm> is the path to the compression executable to
• [end_types] are the ending file types
• The test images are located at the following path: ./img/in/it-#/[start_types]/
• The end images will be created to the following path: ./img/out/it-#/[start_types]/[end_types]/
• The input file info for all input files must exist at ./img/in/it-#/info.csv be run

info.csv Format

The info.csv file should contain the following fields, with one line per input file. A file is considered to be unique if it has a different type than another file of the same image, or is of a different size, so there should be a line for each file within the iteration number directory. "title", "size", "description"

info.csv Generation

The script called info_csv_gen.sh will generate the info CSV for the files in the given file type (other than the image description) so that the user does not have to do all that by hand. Usage is: info_csv_gen.sh <iteration #> <file type> There must be readable files at: ./img/in/it-#/<type>/*.<type>