Low complexity image enhancement GAN-based algorithm for improving low-resolution image crop disease recognition and diagnosis

• Published in: Multimedia tools and applications Vol. 81; no. 6; pp. 8519 - 8538
• Main Authors: Salmi, Abderrahmane, Benierbah, Said, Ghazi, Mehdi
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.03.2022; Springer Nature B.V
• Subjects: Algorithms; Complexity; Computer Communication Networks; Computer Science; Crop diseases; Data Structures and Information Theory; Diagnosis; Generative adversarial networks; Image analysis; Image classification; Image enhancement; Image resolution; Medical imaging; Multimedia Information Systems; Object recognition; Plant diseases; Special Purpose and Application-Based Systems; Smart agriculture; Plant diseases; Image analysis; Super-resolution; Diseases classification
• ISSN: 1380-7501; 1573-7721
• DOI: 10.1007/s11042-022-12256-w

Image analysis plays a crucial role in many real-world applications such as smart agriculture. For plant diseases diagnosis, one of the most recent challenges is the improvement of the plant diseases classification on Low-Resolution (LR) images. The farmer is supposed to obtain High-Resolution (HR) images of plant leaves from the field. Because of the small size of plant leaves and other limitations, the obtained HR images can miss some detailed information that results in blurred LR images of leaves with fewer details. In this paper, we introduce a novel Super-Resolution (SR) algorithm named Wider-activation for Attention-mechanism based on a Generative Adversarial Network (WAGAN) to improve the classification of the tomato diseases LR images. The WAGAN consists of three main parts; the generator network, which has the Wider Activation for Residual Channel Attention Block (WARCAB) as the principal block, the discriminator network and the adversarial loss. To evaluate the potential of the proposed method in plant diseases recognition, we first recovered SR plant diseases images from LR images using the WAGAN. Next, we implemented diseases classification using LR, SR, and HR images. The results proved the efficacy of the proposed method (97.63%) with ×3.6 lower complexity than the state-of-the-art method and very close to the reference HR (97.81%) accuracy. Due to the efficient design of WARCAB and the adversarial loss, the WAGAN focuses more on edges, textures, and other valuable information, which are the key information, needed for the classifier to recognize the disease.