Development of high performance industrial camera and intelligent recognition system for vision-based quality control

Machine vision systems are the whole of the methods and technologies used to perform automated quality control, or analysis, especially on an industrial image processing basis. Today, as the number of facilities which use fast mass production lines increases, the requirements of systems and algorithms which need to be used in such systems are also increasing. A quality control system based on image processing has two components which are industrial digital camera and intelligent recognition algorithm. In this PhD thesis two contributions are presented to develop an image processing based intelligent control algorithm system. The first is a programmable industrial camera which has image construction algorithm and communication algorithm, the second is development of the intelligent recognition algorithms for counting of fast moving objects on a conveyor, the measurement of their dimensions and the detection of possible faults. First of all, the algorithms originally developed in the scope of thesis were developed for a digital camera with programmable resolution in the range of 0.3 MP - 5 MP using Cyclone IV series FPGA and realized in real time to communicate at about 1 Gbps. After, different algorithms which are developed for counting, measuring dimensions and detecting faults of objects that will move rapidly on mass production lines have been verified through different images and objects. Furthermore, it is provided that these developed algorithms can be operated on an interface with the interface design which is developed in the scope of the thesis. The quality control systems, which is obtained within the scope of the thesis, basically contains a digital camera, intelligent control software and main processing unit modules. In conclusion, with the studies done in thesis, some contributions are provided to science with production of image obtaining and digital communication algorithms which are required for development of industrial digital camera in terms of developing the algorithms which are necessary for object counting, dimension measurement and fault detection. The studies developed within the scope of doctoral thesis were supported by the SAN-TEZ project number 0743.STZ.2014 by the Ministry of Science, Industry and Technology.