Study On Front-end System For Smart Camera Based On Zynq

Combined with the functions of image acquisition, processing, analysis and transmission,smart camera has become one of the research hotspots in machine vision. With the features ofminiaturized, distributed and networked structure, it has a broad application prospect indifferent fields. This paper focuses on the front-end image acquisition system for smartcamera based on Zynq platform. The key techniques and implementations of CCD datastream on-chip transmission model, image acquisition system circuit, IP core of CCD driverlogic, image acquisition mode, and performance evaluation are discussed in detail. The maincontributions are as follows:Based on a thorough analysis of the Zynq platform, the overall architecture of smartcamera is provided. The main functional modules including image acquisition, coreprocessing control, Ethernet communication, display, external storage are partitioned. TheCCD data stream on-chip transmission model and XSVI general internal data transmissioninterface are also discussed.The key circuit modules of CCD drive peripheral, AD sampling conversion, CCD drivepower supply are designed for ICX618image sensor. CXD1267AN is used to convert CCDthree-level driving voltage. AD9949is used to handle analogue-to-digital conversion andcontrol signal driving. To achieve general versatility of driving and acquisition mode,finite-state machine is applied to describe the complex driving logic of image acquisition. IPcores of multi-module complex coordination are implemented utilizing Verilog hardwaredescription language. The main function modules include: CCD driving timing of exposure,charge readout, transfer and detecting; SPI configuration protocol of AD registers; XSVIgeneral internal data transmission interface; color restore logic by bilinear interpolation onBayer array.Prototype of image acquisition front-end system for smart camera is developed, withdifferent image acquisition modes implemented, including internal/external trigger, singleframe/continuous acquisition, color/black-and-white image. System driving signal integrity isalso analyzed. CCD performance parameters such as dark current noise, responsivity anddynamic range are evaluated and a preliminary application for optical mark point identification is tested. The results show that the image performance of the acquisition systemmeets the requirements, and the system architecture and implementation also meet thefunctional requirements of smart camera.