Advances in Science, Technology and Engineering Systems Journal Vol. 4, No. 3, 171-182 (2019) www.astesj.com Special Issue on Advancement in Engineering and Computer Science ASTES Journal ISSN: 2415-6698 An Efficient Automotive Paint Defect Detection System Sohail Akhtar * ,1 , Adarsh Tandiya 2 , Medhat Moussa 1 , Cole Tarry 1 1 The Robotic Institute, School of Engineering, University of Guelph, Guelph, ON, N1G 2W1, Canada 2 Praemo Inc., Kitchener, ON, N2G 2Z3, Canada ARTICLEINFO ABSTRACT Article history: Received: 25 March, 2019 Accepted: 23 May, 2019 Online: 12 June, 2019 Keywords: Defect detection Deflectometry Profilometry Automatic inspection Painted surface inspection Vision-based defect detection techniques are widely used for quality con- trol purposes. In this work, an efficient deflectometry based detection system is developed for semi-specular/painted surface defect detection. This system consists of a robotic arm that carries a screen/camera setup and can detect defects on large surfaces with different topologies, such as a car bumper, by traversing its profile. A hybrid pipeline is designed that utilizes multi-threading for optimal resource utilization and pro- cess speed. Specific filters are also designed to remove spurious defects introduced by acute curvature changes and part edges. The system was successful in consistently detecting various defects on small test samples as well as on large bumper parts with varying topology and color and can accommodate inherent ambient lighting and vibration issues. 1 Introduction Quality control is a crucial factor in manufacturing industry as it affects customer satisfaction, reduces production cost and increases profitability. In the auto- motive industry, vehicles are usually assembled from parts shipped by various original equipment manu- facturers (OEM) to the assembly plant. Example of these parts includes front and back bumper covers, side fenders, and other exterior parts which are nor- mally manufactured and painted to specific colour before being shipped to another plant for assembly. These outer body parts are made by Thermoplastic PolyOlefin (TPO) injection moulding process. This process consists of three main steps: moulding, clean- ing, and painting. Defects may induce during any of these processes, which results in part rejection or re- work, causing loss of revenue. As such, it is essential to perform a full inspection of every part. This inspec- tion process is usually carried out by human inspec- tors. It is a costly and a labour intensive job which requires multiple inspection lines for high volume yield. Further, the defect judgement is very subjective, which results in inconsistencies. As a result, overall productivity and quality are diminished. This paper presents a system for inspecting automotive painted semi-specular exterior body parts and is an extension of work originally presented in 15th Conference on Computer and Robot Vision (CRV 2018) [1]. This ex- tended version includes expended testing and analysis. There are a series of challenges that impact the development of an inspection system for this task, in- cluding: 1. Parts are in motion while they are inspected on the production line. The vibrations induced from this motion makes profiling the surface harder. 2. The inspection process must fit within the exist- ing production cycle time. 3. The part being inspected vary in curvature, size, shape and material, with different specular char- acteristics. 4. The visibility characteristic of the defect depends on the external lighting conditions. Designing a proper lighting environment for such a system is a challenging task due to high reflection coeffi- cient of the test surface [2]. 5. Each type of defect varies in shape and size, and experienced inspectors even miss some defects. In recent years, vision-based surface inspection sys- tems have found a burst of application in areas such as defect detection in aluminium sheets [2], locomotive * Corresponding Author: Sohail Akhtar, University of Guelph, 50 Stone Rd. E, Guelph, ON, soakhtar@uoguelph.ca www.astesj.com 171