Benchmarking Image Filtering Methods for Improved Plant Disease Recognition

The agricultural sector plays a vital role in ensuring food security worldwide. Increased agrarian yields satisfy the demands that the world requires. To increase agricultural production, healthy crops without diseases have to be harvested. Nowadays, various kinds of diseases are affecting plants, which cause insecurity in producing the required amount of agricultural yields. To get rid of this earliest identification of diseases is a must. For identifying the diseases that are affecting plants, one of the most important steps known as pre-processing has to be done. This pre-processing step uses various types of filtering techniques to enhance the image quality so that the disease can be easily resolved. In this proposed study, the use of Gaussian filtering, Median filtering, bilateral filtering and Total Variation Denoising (TVD) filtering techniques helps in enhancing the image quality. Further, the study proves that the work of Total Variation Denoising filtering technique works very significantly than the other filters, such as Gaussian filtering, Median filtering, and Bilateral filtering, and is validated by using performance metrics such as MSE, PSNR, and SSIM. Additionally, the study concludes by saying that the Total Variation Denoising filtering technique works dynamically in eradicating the impurities that are associated with the images which results in enhancing the image quality.