Shape feature analysis for different shape detection of computer sketch

Shape feature analysis is one of well-known research topics in image processing and can be implemented in many fields. This paper proposes a method to find a shape which is different from the other shapes of computer sketch. First, the input image will be preprocessed, such as gray scaling and blurring. After that, edges are detected using Canny algorithm and contours of the image are traced. From found contours, sorting and segmentation will be done. Finally, discrete Fourier transform is calculated and dissimilarity measurement with a template will be done using 4 different calculations such as Euclidian distance, cosine similarity, correlation method and Manhattan distance. The goal is to find the most different object. The proposed method is not only trying to deal with simple computer sketch but also more complex computer sketch such as overlapping objects in handwritten computer sketch and noise image. From the research, the most different object can be found in simple image, but it is still hard to be found in complex image. As future task, the research will be improved to deal with more complex image.

I. Introduction

Image processing is one of ongoing research in computer science field. One subtheme of image processing is shape features analysis. Shape features analysis is a process of analyzing geometric shape features which later can be used for shape matching [1]. There are many applications which implement shape feature analysis such as factory product inspection, biomedical engineering, finger print recognition, and content based image retrieval, etc. Shape features analysis is still one of interesting researches in image processing because there are a lot of methods which can be used and different applications which can be created. Human eyes can distinguish different object out of several objects but take a little more time when there are so many objects. It is also a bit harder to distinguish different object when it has similar shape, color or texture with another objects. Computer or robot can help human to make the job easier and faster, by detect that different object automatically.