Image Rectification Software Test Automation Using a Robotic ARM

Abstract

In this paper, we discuss the use of a robotic arm for testing phone software features, such as image rectification, on mobile devices. The problem statement is that we needed an accurate and a precise test automation system for testing and validating the computer vision algorithms used for image rectification in a mobile phone. Manual testing may be error-prone and tedious and thereby the need for a reliable test automation system is of utmost necessity to check the software quality of the image rectification algorithms. The solution to this problem was to design and develop a test automation system using a robotic arm to validate the image rectification algorithms. The robotic arm-based software test automation was deployed and has performed functional performance-based stability tests on multiple software products. The reason for using a robotic arm setup is because it provides us with the flexibility to run our test cases using different speeds, rotation angles, and tilting angles. In this paper, we describe how the robotic arm rotation works. We first measure the center coordinate of the test subject relative to the base of the robotic arm. Then, a 3-D model of the subject is created with those coordinates via simulation mode to represent the real distance ratio setup. Then, the tip of the robotic arm is moved to the proper distance facing the subject. The tests were executed with clear and blurry images containing text with and without image rectification enabled. The result shows the increase in accuracy of text recognition with image rectification algorithm enabled. This paper talks about the design and development of the test automation for the image rectification feature and how we have used a robotic arm for automating this use case.