Leaf RGB image rapid cutting and multiple denoising method

Abstract

The invention discloses a leaf RGB (red, green and blue) image rapid cutting and multiple denoising method, which belongs to the field of leaf color detection and comprises the steps of leaf collection and image acquisition, image cutting and color information extraction, leaf color parameter extraction and denoising to obtain a leaf color image. According to the method, the characteristics of C1-C4 are combined, background difference processing is firstly carried out, and foreground and background are simply separated; the shadow interference of the twisted lead is reduced through edge object removal; a better edge contour is obtained through edge edge recognition, filling, smoothing and multiple times of median filtering; then connected domain area screening is carried out, background miscellaneous points are removed, an accurate foreground target image is separated out, the overall processing effect on monochromatic and mixed-color leaves is optimal, the interference of leaf shadows is further reduced, the processing efficiency is moderate, the parameter accuracy rate is high, and the comprehensive performance is optimal.

Description

technical field [0001] The invention belongs to the technical field of leaf color detection, and in particular relates to a method for rapid cutting and multiple denoising of RGB images of leaves. Background technique [0002] Existing computer automatic cutting methods are mainly realized by MATLAB software. There are mainly C1, color threshold method; C2, OTSU optimal threshold method; C3, edge recognition segmentation method and C4, composite method. However, there are still some problems with this method. [0003] C1. Color threshold method: by pre-setting the color threshold (in this paper, the color RGB of the bottom plate of the sampling platform is used as the threshold) to identify, screen and separate the color of each pixel of the image. [0004] C2. OTSU optimal threshold method: use the im2double function to convert the image into a double-precision array; use the graythresh function to obtain the optimal threshold, and then use the im2bw function to fill in...

AI Technical Summary

Problems solved by technology

[0007] The color threshold algorithm (C1) and OTSU algorithm (C2) can better remove the background of solid-color leaves, but misjudgment will occur for variegated leaves, resulting in a decrease in the accuracy of pixel recognition; while the edge recognition algorithm (C3) can be better The identification of the edge of variegated leaves will not cut redundantly, but the background noise seriously interferes with it, and it is sensitive to leaf shadows, which is easy to cause misjudgment, and its processing takes a long time and is low in efficiency; Denoising is performed again on the basis of edge cutting, which has a better processing effect on both solid-color leaves and variegated leaves, and the extracted leaf color feature parameter values ​​are also better, and the noise-removing effect on leaf shadows is better, but the processing is time-consuming Highest

Images