Shaffer

Overview

Code for ACM MM2016 paper "A Novel Shadow-Free Feature Extractor for Real-Time Road Detection" ( ieeexplore paper html pdf )

This project is part of OpenVehicleVision.

Shaffer takes a color image as input and outputs a grayscale image, where the shadows are disappeared (as shown below). The required parameter, bias, is camera dependent, so we only need to calibrate once for images captured with same camera settings.

Before:

After:
Your browser does not support the canvas tag.

• Select an image: h49-50_1-2.png h53_1-3.png h87-90_1.png
• Choose Filter: grayscale | shaffer (bias = - + )
• To try your own image: , just Click upload or drag and drop an image to this window.
• Bias may be different for different images, e.g. h54-55_1-2.png h64-65_1-3.png.
• Use uncompressed image (*.png) to get better results. Performance is poorer using compressed image (*.jpg) since the loss of color information, e.g. h49_2.png h49_2.jpg.