Mechanical cutting is very important in many industrial applications. Since mechanical cutting often involves contour following, generally, prior trajectory planning is inevitable. However, in some applications the geometric model of the object for cutting is not available. One possible approach to circumventing this difficulty is to employ a visual servoing approach. As a result, in this paper, an eye-to-hand camera mounted on the ceiling is used to obtain the shape information of the object. Moreover, PH splines are designed to represent the obtained object shape. Based on the obtained PH splines, an interpolator is developed to generate the desired feature command for the Image-Based Visual Servoing (IBVS) structure. Another eye-to-hand camera is placed in a position that can have a full view of the object for cutting so as to provide vision feedback for the IBVS structure. Finally, contouring following experiments of several objects with arbitrary shapes have been conducted to validate the effectiveness of the proposed approach.