Abstract:
Leaf Damage Detection and Pencil Drawing Based on Image Processing Techniques
Hitimana Eric Division of Electronics and Information Engineering (Computer Engineering) Advised by Oubong Gwun, PhD
Digital image processing is an area characterized by the need for extensive
experimental work to establish the viability of the proposed solutions to a given
problem. The given problem to deal with can be either in computer vision or
computer graphics applications.
This thesis proposes two methods to handle problems in computer vision application
for pest damage detection and estimation from leaf image and from Computer
graphics, for rendering an automatic color pencil sketch with hatching lines from an
input image.
vii
Image segmentation is the most challenging issue in computer vision applications,
and most difficulties of crops management in agriculture are the lack of appropriate
and accurate methods for detecting the leaf damage for pests’ treatment.
An automatic method for leaf damage detection and severity estimation by
avoiding defoliation is proposed. After enhancing the contrast of the original image
using LUT based gamma correction, the image is processed to remove the
background, and the output leaf is clustered using Fuzzy c-means segmentation
method in V channel of YUV color space to maximize all damages detection, and
finally, the severity damage of leaf is measured in terms of ratio of leaf pixel
distribution between the normal and the detected leaf damage. The results are
compared to the current researches and the accuracy is obvious either in the
background removal or damage detection. From the point of view, the proposed
method can be recommended to famers as a solution of the current naked eye
detection method resulting to errors and time costing.
On the other hand, Non-Photorealistic Rendering (NPR) techniques in computer
graphics to render an image can be created by making use of image editing and
authoring tools. However, these tools are known to bring about high production cost
along with increasing the time and labor charge.
This thesis also proposes an automatic generation of pencil sketch rendition with
hatching lines algorithm. From the input color image, the hatching lines are
viii
generated where their directions depend on the choice of the orientation by motion
blurring; shaded image with contour generated using Kirsch operator is produced to
control the hatching texture on the final pencil sketch result. The algorithm also can
come up with a colored pencil sketch where the color components are got from the
original image.
The method can be used as an editing tool for computer arts, advertising industrial
or technical design, etc
