Thanks Alan.
I really still have a lot to learn. Its fun! Industrial products are the way to go. Today I'm going to turn over my #1 detector over to the dump owner. #5 false signals should not be a problem since there are no #5 bottles. We shall see how it goes in actual operation. If it works OK we will have our first product. I will video the operation. I hope it is not a fiasco. Next going to try my luck at #2 plastic detection. #2 is transparent at terahertz frequencies. I already have the PIR (passive infrared) detector. I am going to try one of those etched plate Edmond Scientific visible spectrum analyzers as a cheep infrared polarizer. heat source----> etched plate---> #2 plastic----->etched plate 90 deg ---> PIR detector We shall see what happens. Maybe I will learn more and advance by two steps. So far the costs have been low and Its been a learning experience. Sold 14 books in Feb. That produced revenue of $2 per day. My cat could live on it. Thank God I had a regular job for 32 years. This starting a business is difficult. Frank -----Original Message----- From: AlanG <a...@magicsound.us> To: vortex-l <vortex-l@eskimo.com> Sent: Fri, Feb 28, 2014 1:45 pm Subject: Re: [Vo]:Plastic detector find On 2/28/2014 6:09 AM, fznidar...@aol.com wrote: Thank you Alan G. How? I really want to do this. I would start with a process-control camera module, maybe 320x240 pixels. You shouldn't need more resolution, and keeping the pixel count small means you won't need a fancy image processing chip. Such modules typically use the common I2C or similar serial control and data interface, and some modules will have an on-board micro controller. If I were doing this project, I'd write some firmware that would look at the change in color information between adjacent pixels and the do some simple statistical analysis on the result. The goal is to generate a single value for each image that represents how many colors are contained in the image. Then all you need is a threshold value above which the #5 plastic is detected. Other thresholds based on luminance might be capable of sorting out the other types so that a single detector could do the whole thing. AlanG