Tracking of Objects Based From Color

 

button3Tracking of Objects Based From Color

By: Mohd Faid Bin Yahya (PhD Student)

 

This article explains about a tracking vision system which emphasizes on objects’ color. The tracking method consists of image acquisition, color space conversion, color thresholding, morphological operations, target detection, and recognition. The flow of the proposed method is shown in Fig. 1.

3a aug15

Fig. 1.Tracking method overview

 

Initially, an image is acquired from a charged-coupled device (CCD) image sensor in the form of a webcam at 15 Hz (15 frames per second). The obtained original image which is in Red-Green-Blue (RGB) color space is converted to Hue-Saturation-Value (HSV) color space for easier image manipulation. Then, color thresholding is used to separate the light sources (targets) from the surrounding [1]. The selected threshold values were 0 to 15 for H component, 150 to 255 for S component, and 246 to 255 for V component. It is important to note that the artificial objects are selected as substitute to light sources since they both can be threshold with the same values of HSV color filter. Thereafter, morphological opening is utilized to remove noises while preserving the shape and size of the targets. Subsequently, for target detection, there are 3 steps required known as contour finding, contour moment calculation, and contour filtering. Finally, the recognition of the target is based from evaluating positions of centroid of these detected contours/targets. Fig. 2 shows targets tracking at different distances.

 

3b aug15

 

Fig. 2. Image captured at different distances

 

Reference(s):

  1. E.R. Davies (2010). Computer and Machine Vision: Theory, Algorithms, Practicalities. 4th Edition, Elsevier