SoC Based SIFT Algorithm for Identification of the Quality Objects for Palletization Application

Abstract

In some of the manufacturing industries automation problems occur most commonly like quality control and palletization. This paper proposes a system level design for the identification of quality objects for palletization process to solve the automation problems. The quality of the object is identified by using SIFT (Scale Invariant Feature Transform) algorithm which is one of the image processing algorithms most popularly used for local feature detection. By using SIFT algorithm featurepoint extractions are performed for an input image. After featurepoint extraction, feature matching is performed by comparing the extracted featurepoints of the input image and original image featurepoints which are stored in database. The output comes after the feature matching will actuate the palletization process to differentiate the type of objects. The processing of an image using SIFT algorithm for palletization process is designed in MATLAB by using Xilinx System Generator and simulations are performed. Then the HDL code is generated for the whole design by using system generator. The generated HDL code is synthesized by Design vision compiler in Synopsys tool. To integrate the whole system design into SoC (System on Chip) using 90nm technology physical design is performed in Cadence Encounter tool. Due to this design the whole system obtains optimized area and power.