This Assemblyline service provides image analysis.
NOTE: This service does not require you to buy any licence and is preinstalled and working after a default installation
File types currently supported:
- image/*
This program is a Optical Character Recognition (OCR) engine, which attempts to extract text from images
Tesseract is licensed under the Apache License, Version 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
Source code is found here: https://github.com/tesseract-ocr/tesseract
AL outputs:
- Text extracted and appended to file "output.txt"
In this service, you're allowed to override the default OCR terms from the service base using ocr
key in the config
block of the service manifest.
Let's say, I want to use a custom set of terms for ransomware
detection. Then I can set the following:
config:
ocr:
ransomware: ['bad1', 'bad2', ...]
This will cause the service to only use the terms I've specified when looking for ransomware
terms. This is still subject to the hit threshold defined in the service base.
Let's say, I want to use a custom set of terms for ransomware
detection and I want to set the hit threshold to 1
instead of 2
(default). Then I can set the following:
config:
ocr:
ransomware:
terms: ['bad1', 'bad2', ...]
threshold: 1
This will cause the service to only use the terms I've specified when looking for ransomware
terms and is subject to the hit threshold I've defined.
Let's say, I want to add/remove a set of terms from the default set for ransomware
detection. Then I can set the following:
config:
ocr:
ransomware:
include: ['bad1', 'bad2', ...]
exclude: ['bank account']
This will cause the service to add the terms listed in include
and remove the terms in exclude
when looking for ransomware
terms in OCR detection with the default set.
Please note that modules are optional (see service configuration). They are provided for academic purposes, and are not considered ready for production environments
Current AL modules:
Least significant bit (LSB) analysis:
-
Visual attack
-
Chi square
-
LSB averages (idea from: http://guillermito2.net/stegano/tools/)
-
Couples analysis (python code created largely from java code found here: https://github.com/b3dk7/StegExpose/blob/master/SamplePairs.java)