A segmentation method of bagged green apple image

• Published in: Scientia horticulturae 2019-02, Vol.246, p.411-417
• Main Authors: Lv, Jidong, Wang, Fan, Xu, Liming, Ma, Zhenghua, Yang, Biao
• Format: Article
• Language: English
• Subjects: Apple; Image enhancement; Image reconstruction; Image segmentation
• ISSN: 0304-4238; 1879-1018
• DOI: 10.1016/j.scienta.2018.11.030

•A segmentation method of merging after regions extraction for bagged apple image.•Normal-light fruit area oriented extraction based on CLAHE method.•Highlighted fruit area oriented extraction based on image reconstruction. Based on the own characteristics of bagged green apple image, the work designed an image segmentation method of bagged green apples which was combined after extraction of the normal light regions and highlighted regions of fruits. The contrast limited adaptive histogram equalization (CLAHE) algorithm was deepened fruit green region and improve edge definition in image, and then the R-B color difference image based on CLAHE image was obtained. The linear contrast of the original RGB image was enhanced to improve color difference in the normal light regions of fruits and leaves. The image after contrast enhancement was extracted in the region of R-B color difference image. Then the image was performed with OTSU segmentation algorithm and denoising to extract the normal light region of fruit. For the extraction of highlighted region of fruit in RGB images of bagged green apples, the main colors of image were first extracted. Then, the image was reconstructed with the main colors. The reconstructed and original images were subtracted and denoised to realize the extraction of the highlighted region of fruits. Finally, the complete fruit target region was obtained by combining the two extracted regions. The test results showed that we could obtain more complete fruit region of bagged green fruits based on the method in the work.