Cross-Browser Testing in Browserbite

Abstract

Cross-browser compatibility testing aims at verifying that a web page is rendered as intended by its developers across multiple browsers and platforms. Browserbite is a tool for cross-browser testing based on comparison of screenshots with the aim of identifying differences that a user may perceive as incompatibilities. Browserbite is based on segmentation and image comparison techniques adapted from the field of computer vision. The key idea is to first extract web page regions via segmentation and then to match and compare these regions pairwise based on geometry and pixel density distribution. Additional accuracy is achieved by post-processing the output of the region comparison step via supervised machine learning techniques. In this way, compatibility checking is performed based purely on screenshots rather than relying on the Document Object Model (DOM), an alternative that often leads to missed incompatibilities. Detected incompatibilities in Browserbite are overlaid on top of screenshots in order to assist users during cross-browser testing.