Segmentation of color images for image data analysis
Sanmati Kamath · 2006
The work described here discusses the analysis of images in port settings. As cargo containers are of specific colors, the hue information is being exploited to segment, identify and compress such video streams. A novel scheme using multiple processes has been developed. The first process is obtaining the vector angle between the RGB components which is used to segment images based on the color. The second process is the Hough transformation that is carried out on the vector angle result as well as a Laplacian of the saturation image. Using the Hough transform result on both the vector angle result and the Laplacian, various broken segments are connected effectively without adding redundancy. This information can then be used to detect corners and perform compression.