Design of Multi-Level Color Thresholding and Geometric Object Orientation Estimation Algorithms for Robotic Vision Systems

• Main Authors: Tsung-Yen Liu, 劉宗諺
• Other Authors: Chi-Yi Tsai
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/41734241675560302746

碩士 === 淡江大學 === 電機工程學系碩士班 === 101 === Color image segmentation is one of the most important preliminary tasks in robotic vision systems. This thesis presents a novel unsupervised multilevel color thresholding algorithm to address this issue efficiently. The proposed algorithm consists of a learning process and a multi-threshold searching process. The former aims to learn the color distribution of an input video sequence in HSV color space, and the latter automatically determines the optimal multiple thresholds to segment all colors-of-interest in the video based on a new class-variance criterion. In the learn process, a novel color-distribution learning algorithm cooperating with a color-pixel extraction method is proposed to learn a color distribution model of all colors-of-interest in the video images. In the multi-threshold searching process, a nonparametric multilevel color thresholding algorithm with an extended within-class variance criterion is proposed to automatically find the optimal upper-bound and lower-bound threshold values of each color channel. After segmenting objects-of-interest, this thesis also proposes an image-based object orientation estimation algorithm, which is developed based on the projection of a geometric object with different angles to accurately and efficiently estimate its orientation from a single view image. Simulation and experimental results show that both proposed algorithms not only provide satisfactory results, but also are suitable to combine with a robot manipulator system for achieving object-segmentation and pick-and-place tasks.