Product classification analysis method based on binary density clustering

Abstract

The invention discloses a product classification analysis method based on binary density clustering. The method comprises the following steps: obtaining an initial product waveform data set; performing preprocessing operation to obtain a to-be-analyzed product waveform data set, and performing modulation interval to realize single wave division; performing density clustering analysis on each single wave in the multi-dimensional array according to wave intensity to obtain a single wave intensity coding array, and executing dynamic time warping operation on the single wave intensity coding array; performing statistical counting on each single wave in the multi-dimensional array according to the duration of each intensity of the wave to obtain a single wave duration array, and performing dynamic time warping operation on the single wave duration array; and respectively comparing and classifying the single wave time similarity and the strength similarity with a similarity threshold value, and counting the type number of the waveforms and the total quantity of the waves under the same waveform. According to the invention, all products on one production line can be analyzed in a unified manner, the types and the number of the products on the production line can be identified, and the total product quantity of each type of products can be obtained through accurate analysis.

Description

technical field [0001] The invention relates to a product classification algorithm, in particular to a product classification analysis method based on binary density clustering. Background technique [0002] In the existing technology, the degree of automation of the industrial production line has received more and more attention, but a large amount of production data cannot be effectively analyzed and used. For example, the existing product identification technology mainly includes two technologies based on labels and machine vision. , where label technology includes barcode identification and RFID electronic label technology, RFID electronic label technology is not widely popular due to cost and other issues, while machine vision technology does not need to add additional labels to products, and can identify products according to the appearance of product packaging, which is suitable for packaging Products with rich appearance and texture, however, are detected in real tim...

AI Technical Summary

Problems solved by technology

[0002] In the existing technology, the degree of automation of the industrial production line has received more and more attention, but a large amount of production data cannot be effectively analyzed and used. For example, the existing product identification technology mainly includes two technologies based on labels and machine vision. , where label technology includes barcode identification and RFID electronic label technology, RFID electronic label technology is not widely popular due to cost and other issues, while machine vision technology does not need to add additional labels to products, and can identify products according to the appearance of product packaging, which is suitable for packaging Products with rich appearance and texture, but the detection is real-time detection, which is sensitive to changes in external lighting and shape deformation, which is not conducive to product classification and analysis with low cost and small computing overhead

Images