A locomotive is a rail vehicle that provides the driving power for a train. Railway vehicles consist of multiple components, among which bogies play a critical role as key load-bearing parts. They support the weight of the vehicle body and mitigate the impact of track irregularities to ensure smooth and safe operation. To achieve a lightweight assembly, bogie frames are typically constructed using welded structures. Welding is one of the most crucial methods in modern industrial manufacturing and is widely employed in the design of lightweight structures. However, during the welding process, major defects such as weld deformation and residual stresses are inevitable, which can affect assembly accuracy and increase production costs. Traditionally, this welding process is carried out manually, requiring high operator expertise and significant time to execute. This paper presents a welding path planning algorithm aimed at achieving higher welding quality, reducing the overall welding process time, eliminating the production process's dependence on the skills of specific operators, and enhancing the precision in manufacturing various structures used in the rail industry, such as locomotive bogies. The use of image processing for the identification and path planning of welding joints reduces the high costs associated with procuring online laser tracking devices. Moreover, this paper introduces a comprehensive method for processing the entire image and providing the welding path for offline robot programming without any prior knowledge of the joint location. Automatically and accurately determining the seam locations via the robot represents a significant advancement toward automating the arc welding process. The proposed methods have been tested and validated using a laboratory prototype of a 2DOF planar welding robot. The results confirm an accuracy of 2 mm for parabolic joints and 3 mm for straight joints, meeting the precision requirements of small and medium-sized industries.