Surface defects represent a major threat for product quality and its function that require proper inspection. Variety of surface defects makes their inspection more complicated, costly and requires longer time. Reliance on human inspection can lead also to less consistent results due to the variance in expertise and human error. For those reasons, traditional inspection methods less fit to fast automated manufacturing systems. Employing computer vision techniques in vision-based Inspection systems (VBI) can lead to developing better systems that match modern manufacturing systems in terms of speed, automation, higher productivity, less dependency human experience and cost optimization. In this research, an automated vison-based inspection system (CAI-2) is developed for detection and classification of surface defects encountered in metal parts using Digital Image Processing (DIP) techniques. CAI-2 receives the image of the part under inspection as an input, detects and generates automatically the type, number and location of existing surface defects. Six types of surface can be detected using the proposed method including Cracks, dents, fretting, flaking, rust, and smearing. The accuracy and effectiveness of the developed model were evaluated against skilled inspectors by measuring the values of inspection time, recall, precision and f-measure parameters values. Experimental results proved competitive accuracy and efficiency of the proposed inspection model compared to traditional inspection methods.