| Abstract:At present, leaf feature recognition methods mostly recognize leaf shape, vein and other features according to leaf color, resulting in the problem of low definition of the recognized image features. A dynamic recognition method of tobacco leaf image frame features with production line quality traceability is proposed. The wireless network hard disk video recorder is used to set up the leaf image acquisition process, collect the tobacco leaf image during the quality traceability of the production line, and introduce the histogram equalization, grayscale and noise reduction preprocessing technology to preprocess the leaf image. The quadtree splitting algorithm is introduced to segment the tobacco leaf image. Starting from the leaf texture, through the extraction technology of fractal dimension, gap amount, energy, entropy, contrast, contrast and correlation features, the characteristics of leaf image frame texture complexity, density, gray level, amount of information, clarity and linear correlation are extracted. According to the results of image frame feature extraction, DS synthesis algorithm is used to dynamically recognize the frame features of tobacco leaf image. Result shows that after the application of this research method, the vein feature clarity of cigar tobacco leaves was improved by 8.2 and 2.68 respectively. The clarity of vein characteristics of burley tobacco leaves increased by 7.56 and 1.5 respectively. |
点此下载全文
