Submitted to: ASAE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: 8/15/1999
Publication Date: N/A
Citation: Pelletier, M.G., Barker, G.L., Baker, Jr., R.V. Non-contact image processing in gin trash sensors in stripper harvested cotton. 1999. American Society of Agricultural Engineers. Paper No. 991126. 9 p.
Interpretive Summary: The research consists of a real-time computer recognition system for seed cotton trash or cotton lint trash in situ, without the need for any mechanical sample preparation or presentation. The technique involves acquiring an image of seed cotton in a free-form, as would occur while the cotton lays naturally upon a conveying belt, in a cotton module, or in an alternate, preferred location. The technique measures the image area coverage of trash, lint or seed cotton, and distinguishes each from the background, moving the influence of voids from the measurements. After classifying the image, it determines the percent trash content and then classifies the trash into the three separate classes of burs, sticks, and leaves. This information is then used to determine the trash content (on a mass basis) of the sample. This system could be used for improving trash recognition in gin process control by removing the constraint of paddle samplers. This will allow trash measurements to be used in new, innovative locations as well as improving the sampling techniques currently in use. There are several needs for trash measurement in cotton ginning processes. The trash content percentage can be used to optimize the seed cotton and lint cleaning processes and help to control the seed cotton drying process. The commercial potential for this technique lies in the application of two to six units in approximately 1200 cotton gins in the United States. It is inevitable that the cotton gins in the near future will become fully automated. This is due to the fact that optimal control of the gin will produce optimal economic returns.
Technical Abstract: This study was initiated to provide the basis for obtaining on-line information as to the levels of the various types of gin trash. The objective is to provide the ginner with knowledge of the quantity of the various trash components in the raw uncleaned seed cotton. This information is currently not available to the ginner for use in optimizing the gin machinery. An existing Kodak trilinear array color ccd line scan imager wa connected to a PC in a laboratory environment. Due to the high levels of trash in stripper harvested cotton, an 8.0 in. by 10.0 in. viewing area was used. Images of seed cotton (taken from various stages in the pre-ginning cleaning process) were obtained without pressing the seed cotton against glass plates. This omission, of a standard cotton image acquisition technique, increases the opportunities for image acquisition in the gin to obtain the gin trash levels for use in optimal gin control by removing the necessity of capturing a sample of seed cotton to press against an imaging plate. Algorithms were developed to differentiate between: seed cotton, the various trash components and the background. Once the individual components were identified, an algorithm was developed to determine the levels of the various trash components; sticks, burrs, and leaves.