The Recognition Count And Monitoring For Product In Production Line Based On ARM

Traditional methods of counting the production line products generally rely on pulsecount, and sometimes due to improper spacing difference there exists the undetectedphenomenon. Some of the mixed product lines currently also exists the phenomenon of falseconsciousness undetected because of the quantity, model number, and volume sizes. Suchproblems of artificial labor intensive, management expenses and the lack of real timemonitoring of production line counting system cause huge losses to the production.For these reasons, the design of real time counting, recognizing and monitoring systembased on the image acquisition module and the retrieval module is proposed, Solving themixed product line count monitoring and management. Samsung S3C2440is used as the mainchip in the slave computer and the hardware modules design the acquisition system of slavecomputer based on embedded ARM. First, pull the software environmental configuration intothe slave computer system and load hardware drivers of the image capture device, and thenuse NFS root file system for mounting so that the system can perform networkcommunications and host-compute, and by running the root directory software to captureimages and video capture, and send the pictures saved to the host computer. The retrievalalgorithm of the host computer takes the texture feature as a retrieval feature of retrievalimage, and for retrieval algorithms facing with the whole image rather than the local area,take a algorithm which firstly does picture segmentation, then weighting the search results. Inthe image feature extraction process errors comes out when a slight blurred image on thesystem is retrieved, thus conceiving an algorithm which is firstly restoring fuzzy image thenprocessing retrieval algorithm. Based on this issue, firstly adopt several classic fuzzy imagerestoration methods to recover, and then compare several pictures restored by the classicalgorithm in time and clarity. The projection iterative method is better than the otheralgorithms in terms of time and clarity, so the final blurred image restoration is applied. Thecounting, recognizing and monitoring system of production line is researched based on thecombination of above software and hardware components discussed, and the construction ofslave computer and choosing the capture device is the architecture foundation of the hardwarepart. The image retrieval algorithms and acquisition programs are the soul to achieve thesystem function, and the algorithm processing section and the processor are the key to impacton the real time of the system. After the images captured by slave-computer are sent to thehost computer, the image processing algorithms in the host-computer include the restoration of the blurred image, the extraction of image feature, similarity analysis, and giving the countresults.The experiment selects the white as background color and transmission speed0.5m/spaper strip, experiment car drag the paper strip, and adjusts the best light intensity and thecamera position, placing electronic components of different models in it. The slave computercaptures images and video of electronic components, which are sent to the host-computerthrough the Ethernet port. Do image processing algorithms on VC++2008, and experimentalresults show that in the browser real-time videos play well and counting results are accurate,but when the light intensity of the conveyor belt is too strong or too dark, the ambiguity ofimages captured by the camera is larger and the counting results take on bigger error.