Fusion of clahe-based image enhancement with fuzzy set theory on field image

Field image can be defined as image captured from handphone or any mobile device in open or outdoor environment. Field image is also known as low quality, low resolution, noise and affected background. In contrast, image captured in lab or studio is a high-quality image taken in a proper setup using high specification device. In agriculture, field leaf image is commonly used to identify plant disease. Accurate detection of plant disease is needed to strengthen the field of agriculture and economy of the country. The disadvantages of field leaf image, are low resolution, low contrast, blur and unsharp due to inconsistent setting or environment exposures. Image enhancement method helps to improve image quality, reduce impulsive noise, and sharpen the edges of field leaf image. In this study, measurement of contrast level is used to compare the quality between field leaf image and image taken in the studio (lab image). High quality image has high contrast value and it shows that lab image has high contrast value. Therefore, this research is focus on field leaf image enhancement to improve the quality of the image and make it as same quality as lab image. This research presents a framework of fusion techniques namely, Contrast-Limited Adaptive Histogram Equalization (CLAHE), Unsharp Masking (USM) and Fuzzy theory. CLAHE-based and USM image enhancement techniques are widely used to enhance and sharpen the edge of field leaf image. However, the drawback of these techniques is the field leaf image is still in low contrast and not as same quality as the lab image. To further improve the quality of field leaf image, combine the existing framework with Fuzzy Set Theory. Furthermore, there are significant difference when applying the framework in global and local images. Therefore, comparison the performance of the framework is done between global and local images. The result of the proposed image enhancement framework is compared with the lab image as a benchmark. From the results shows that the proposed image enhancement framework produces better quality of field leaf image and required minimum processing time. The evaluation measurement methods used in this research are Contrast Value, Contrast Difference (DC), Contrast Improvement Index (CII) and Peak- Signal-Noise-Ratio (PSNR). The proposed fusion framework proved that field leaf image produces better quality image where the CII value increased from 86% to 94%. It also shows that local-based image enhancement with 4x4 patches produce better quality from global-based image enhancement.

Text ELMALIANA BINTI ALBAHARI - IR.pdf Download (1MB)

Actions (login required)

View Item