Machine Vision: Ping-Pong Ball and Bat Tracking - Project Solution

This document presents a comprehensive solution to a machine vision project focused on tracking a ping-pong ball and bat within a video sequence. The solution utilizes MATLAB and covers several key aspects of image processing and computer vision. It begins with the creation of a custom function, 'createMask.m', using the colorThresholder application to identify the bat and ball through color segmentation in the L*a*b color space. The solution details the process of loading images, selecting color spaces, and refining segmentation using histogram controls. Furthermore, the document addresses object labeling using the bwlabel function and discusses pixel connectivity (4 and 8). It then employs the regionprop function to identify objects based on properties like orientation, major/minor axis length, and centroid, and uses a for loop to label objects with their corresponding labels. The solution also explores threshold-based color segmentation, including two-pass thresholding for improved accuracy. This includes adapting the ROI (Region of Interest) dynamically based on object detection and the use of ROI scaling techniques. Finally, the document presents results, including a visualization of the ball tracking system and its performance analysis, with detection rates provided for different video sequences.