A fast detection method for fruit surface defects based on regional brightness adaptive correction

Abstract

The invention relates to a fast detection method for fruit surface defects with self-adaptive correction of regional brightness. Firstly, a fruit RGB color image is obtained with a black background, and then the background is removed and the R-B difference grayscale image is extracted to form a target image P(x, y), and then use the average of the largest gray values ​​in the neighborhood of each pixel in the image as the brightness of the current pixel to calculate and extract the surface brightness image I(x,y) of the target image P(x,y), and P (x, y) and I(x, y) points are divided to obtain the brightness correction image F(x, y), and the global single threshold method is used to extract the target area for F(x, y) to obtain the target binarized image B(x, y). y), performing area threshold filtering on B(x, y) to obtain the fruit surface defect area image D(x, y). The algorithm of the invention is simple, and the detection of an image can be completed in tens of milliseconds on an ordinary computer, with an accuracy rate of 94.6%, and the image processing time of online fruit detection can be greatly shortened. The invention has high adaptability, low cost, simple operation and good detection robustness to samples with different types of defects.

Description

technical field [0001] The invention belongs to the technical field of computer vision image processing, and in particular relates to a fast detection method for fruit surface defects with area brightness self-adaptive correction. Background technique [0002] Fruit surface defects have always been a difficult point in all external quality inspections of fruits, and are also one of the important basis for fruit grading. When a large number of researchers at home and abroad detect surface defects of fruits through computer vision, they find that most of the fruits are spherical or spheroidal, resulting in uneven reflection of light, resulting in uneven brightness of the fruit surface, which appears as high brightness in the middle area on the grayscale image. , the gray level of the edge area is low, and the defect area usually exists in the form of low gray value, and the gray level of the defect area intersects with the gray level of the normal area, which directly leads to...

AI Technical Summary

Problems solved by technology

[0009] In order to solve the problems of complex fruit surface defect detection algorithm in the background technology, limited detection surface defect types, high hardware cost and difficulty in adapting to online detection requirements, etc., the purpose of the present invention is to provide a fast fruit surface defect detection method based on regional brightness adaptive correction , suitable for online detection occasions

Images