Video and Image Processing Blockset 2.8

Chroma-based Road Tracking

This demo uses chroma information to detect and track road edges set in primarily residential settings where lane markings may not be present. The Chroma-based Tracking demo illustrates how to use the Color Space Conversion block, the Hough Transform block, and the Kalman Filter block to detect and track information using hue and saturation.

Contents

Demo Model

The following figure shows the Chroma-based Tracking model:

Algorithm

The demo algorithm performs a search to define the left and right edges of a road by analyzing video images for change in color behavior. First a search for edge pixels, or a line passing through enough number of chroma pixels, whichever comes first, is initiated from the bottom center of the image. The search moves to both the upper left and right corners of the image.

To process low quality video sequences, where road sides might be difficult to see, or are obstructed, the algorithm will wait for multiple frames of valid edge information. The demo uses the same process to decide when to begin to ignore a side.

Tracking Results

The Detection window shows the road sides detected in the current video frame.

When no road sides are visible, the Tracking window displays an error symbol.

When only one side of the road is visible, the demo displays an arrow parallel to the road side. The direction of the arrow is toward the upper point of intersection between the road side and image boundary.

When both of the road sides are visible, the demo shows an arrow in the center of the road in the direction calculated by averaging the directions of the left and right sides.