Automated Garbage Bag Sorting Combination of powerful real-time image processing and true-colour line-scan cameras offer new colour based opportunities to a wide scale of applications from identification, sorting and grading to real-time process control. One of the new ‘exotic’ machine vision applications is an automated sorting system for household garbage bags. Processing of household waste, and reduction of garbage collection and handling costs are important questions both from the environmental and cost viewpoint. In Finland, in a not densely populated area live 200 000 people. It is estimated that the quantity of household waste they create annually is 68 000 metric tons. 24 000 tons of this can be used for energy production and 15 000 tons are bio-degradable. Additionally the local industry can deliver 18 000 tons of waste that can be used for energy production. Today the families living in such an area pack their garbage in bags of different colour based on the type of waste. Thus all garbage can be collected and transported in one truck. The separation of bags is done automatically at a centralised sorting and recycling plant. Dry waste that can be used for energy production is packed in yellow or orange bags, bio-degradable waste in green bags, and non-recyclable household garbage in bags of any other colour or mixture of colours, Figure 1. Because a machine vision sorting system can distinguish millions of different colours, the sorting into various categories has no practical limit and bag colours can used very flexibly. The benefits of a new type of garbage collection and processing are:
Figure 1. Garbage bags to be sorted entering in the plant.
Figure 2. Optical systems are at the red ‘trouser-leg’ housings. An optical true-colour sorting system has practically an unlimited sorting capacity, Figure 2. The mechanical sorting, however, limits the processing capacity to 150 garbage bags per minute per conveyor. One of such sorting systems is presented in the presented figures. The system is installed at the recycling plant in Anjalankoski in southern Finland. In Figure 2 are shown the two measuring systems and the four conveyors. Each machine vision system controls two separate conveyors. The red ‘trouser-legs’, that keep the optical path clear, show the measuring locations.
Figure 3. Sorting mechanics in action.
Figure 4. Sorted bags for energy production. Based on the colour analysis of each bag the two-way kickers push the analysed bags to an exit conveyor, Figure 3. The sorted bags are finally stored in a container for transportation to a power plant for energy production, Figure 4. Processing of the image and control of the process takes place in an intelligent, ultra-fast, multi-channel processing system, which has been designed especially for complex, real-time machine vision applications. The processor has a 12-bit dynamic range with a calculation speed of 1000 Mbytes and calculation power up to 46080 Mflops. With the scaleable modular structure it is possible to link 256 cameras together. The data is processed at the site of measurement, where all required calculations, decisions and control commands are executed. Only the finished data and information is sent forward when required. Therefore the system instrumentation is easy and cabling simple and reliable. With the modular DSP design it is easy to create either a one CCD camera system or to design a multi-camera system with practically unlimited number of cameras. The small size, extensive communication ports and easy adaptability of the processor make it an ideal component for all kinds of measurement and process control applications in various industrial applications. Contact: inX Systems Ltd
.
|
