Reduction of reflection and defect detection for metallic jobs using Superresolution Method on Hardware platform

Abstract

The quantitative inspection of metallic jobs is very crucial in manufacturing industries as not only the conventional approach of inspection can’t achieve true outcome but the reflection of light from the metallic jobs also create the hurdles for proper inspection of defects in metallic jobs. Hence in this paper the automatic inspection with the help of image processing have been proposed for better inspection. The hardware friendly Raspberry Pi development board has been utilized for the automation using superresolution (SR) technique. The prototype implementation of SR method that achieves reflection reduction and detection of defects in the metallic jobs independently is implemented. The outer and inner defects of metallic jobs have identified along with reflection reduction. The prototype is implemented for this purpose with the help of image processing superresolution algorithm using DICWF filter bank wavelet transform to obtain superresolution reconstruction on real time Raspberri Pi development board. The coefficients of DICWF are dyadic and integer in nature that can be easily implemented on real time processor.  Performance of the proposed real time system using DICWF for superresolution reconstruction has been evaluated based on subjective and objective results.