Color Recognition of Recycled Glass Using Peak Feature of Color Histogram

Recycling of waste glass is an important means to protect environment and save resources. In the process of recycling, different colors of glass need to be sorted to achieve efficient and full utilization. At present, machine vision is mainly used for glass color recognition, but only a certain color or colors with significant differences can be recognized, while the recognition accuracy for similar colors is low. According to the above problem, a glass color recognition method based on the peak feature of the color histogram of the glass sample transmission image was proposed in this paper. The peak positions and peak separations of the histogram fitting curves of the blue, green and red channels of the transmission image were used as color features, based on which a color recognition model was established by using support vector machine classification algorithm. The test results showed that the recognition accuracy of 8 glass colors, that is colorless, light green, dark green, yellow green, light blue, dark blue, tawny and brown, reached 100%.