Nursuriati Jamil

Automated Grading of Palm Oil Fresh Fruit Bunches (FFB) Using Neuro-fuzzy Technique

Automated fruit grading in local fruit industries are gradually receiving attention as the use of technology in upgrading the quality of food products are now acknowledged. In this paper, outer surface colors of palm oil fresh fruit bunches (FFB) are analyzed to automatically grade the fruits into over ripe, ripe and unripe. We compared two methods of color grading: 1) using RGB digital numbers and 2) colors classifications trained using a supervised learning Hebb technique and graded using fuzzy logic. A total of 90 images are used as the training images and 45 images are tested in the grading process. Overall, automated grading using RGB digital numbers produced an average of 49% success rate, while the neuro-fuzzy approach achieved an accuracy level of 73.3%.

Texture feature extraction using 2-D Gabor Filters

Texture feature extraction is a procedure of computing and describing the features and characteristics of image which numerically describes that texture image properties. This paper investigated texture feature extraction using 2-D Gabor Filter to extract the texture features of Inverse Fast Fourier Transform (IFFT), texture energy and transformed IFFT. The Gabor filter bank experimented on seventy two collected samples of skull-stripped T1-weighted, T2-weighted and FLAIR MRI brain images utilizing four frequencies and four orientations. Results showed that texture feature extractions of two highest frequencies with all four orientations produced the highest acceptance rate.

Skull stripping of MRI brain images using mathematical morphology

Skull stripping is a major phase in MRI brain imaging applications and it refers to the removal of its non-cerebral tissues. The main problem in skull-stripping is the segmentation of the non-cerebral and the intracranial tissues due to their homogeneity intensities. As morphology requires prior binarization of the image, this paper proposed mathematical morphology segmentation using double and Otsus thresholding. The purpose is to identify robust threshold values to remove the non-cerebral tissue from MRI brain images. Ninety collected samples of T1-weighted, T2-weighted and FLAIR MRI brain images are used in the experiments. The results showed promising use of double threholding as a robust threshold value in handling intensity inhomogeneities compared to Otsus thresholding.

Noise removal and enhancement of binary images using morphological operations

Mathematical morphological operations are commonly used as a tool in image processing for extracting image components that are useful in the representation and description of region shape. In this paper, six basic morphological operations are investigated to remove noise and enhance the appearance of binary images. Dilation, erosion, opening, closing, fill and majority operations are tested on twenty-five images and subjectively evaluated based on perceived quality of the enhanced images. Results of the experiments showed that noise can be effectively removed from binary images using combinations of erode-dilate operations. Also, the binary images are significantly enhanced using combinations of majority-close operations.

Image Retrieval of Songket Motifs Using Simple Shape Descriptors

Songket is a traditional handwoven cloth of the Malays and its beauty lies on the design of the songket motifs intricately woven on the cloth. Even though color is a feature of the motif, shape is more important in determining the type of the motifs. This paper describes and tests the efficiency of retrieving songket motifs images using different combination of five geometric shape descriptors. Even though there are many existing approaches of shape-based image retrievals, the goal of this paper is to take advantage of the faster calculations of the geometric shape descriptors to retrieve the images. Fifty selected songket motifs are used as sample queries in this paper. Similarities of these images are measured using Euclidean distance and performance of the retrieval effectiveness is evaluated using recall and precision rate. Results of the experiment showed that geometric shape descriptors are effective in retrieving the songket motifs

A comparison of noise removal techniques in songket motif images

Noise removal or noise suppression is important as the results of the noise removal task have a strong influence on the quality of the following image processing techniques. In this paper, three popular filters that is mean, median and adaptive filters are tested on 25 images of songket motifs. Different sizes of the filter kernels are also used as it plays a factor in producing the desired results. Prior to noise removal, the images are subjected to preprocessing operations such as histogram stretching and image cleaning. To determine the performance of the noise removal, visual evaluation is performed.

A rule-based segmentation method for fruit images under natural illumination

Image segmentation is a process that significantly important for machine vision system such as automatic fruit grading system. This process separates an image into several areas to extract the interest object from its background. However, the segmentation task is difficult for isolating the images that captured in outdoor environment. This is due to the existence of non-uniform illumination on the object surface. Technically, different illuminations lead to different intensity on the object surface colour. This condition leads to low quality segmented images and therefore reduces the accuracy of object classification. Image segmentation can be accomplished using several methods such as Otsu, K-means and Fuzzy C-means. However, these three traditional methods have limitations in producing accurate segmented areas due to the existence of illumination on the object surface. Therefore, this paper developed a rule-based segmentation method that is able to segment natural images correctly and accurately. This method uses IF-THEN algorithm to segment the images of interest object. All four segmentation methods are implemented on fruit images and their performance are compared based on visual and quantitative evaluations. The analysis results showed that the new method is capable to produce segmented images with high accuracy rate.

Gradient-Based Edge Detection of Songket Motifs

This paper discussed the effectiveness of several popular gradient-based edge detection techniques when applied on binary images of songket motifs. Five edge detector algorithms that is Roberts, Sobel, Prewitt, Laplacian of Gaussian and Canny are applied to twenty-five Malaysian traditional songket motifs. These scanned motif images are initially preprocessed to remove noise using several morphological operations. Other than noise removal, binarization of color images are also done to produce binary images. To determine the performance of the edge detectors, the results are evaluated by five human subjects based on several pre-conceived criteria.

An accurate thresholding-based segmentation technique for natural images

Segmentation is a process of dividing an image into distinct regions with the aim to extracts object of interest from the background. The traditional thresholding and clustering segmentation techniques that were widely used are Otsu and K-means, respectively. However, the segmentation process becomes more challenging for segmenting natural images. Both Otsu and K-means methods failed to produce good quality of segmented areas under natural environment due to the complex background and non-uniform illumination on the images. Therefore, this paper proposed an improved thresholding-based segmentation with inverse technique (TsTN) that was able to partition natural images. A comparison between Otsu, K-means and TsTN techniques was conducted using colour-based image processes on the quality of the segmented images. The analysis results showed that TsTN has the ability to produce good quality of segmented images. Furthermore, this improved technique was proved to be more accurate than the traditional thresholding and clustering techniques.

A modified edge-based region growing segmentation of geometric objects

Region growing and edge detection are two popular and common techniques used for image segmentation. Region growing is preferred over edge detection methods because it is more robust against low contrast problems and effectively addresses the connectivity issues faced by edge detectors. Edgebased techniques, on the other hand, can significantly reduce useless information while preserving the important structural properties in an image. Recent studies have shown that combining region growing and edge methods for segmentation will produce much better results. This paper proposed using edge information to automatically select seed pixels and guide the process of region growing in segmenting geometric objects from an image. The geometric objects are songket motifs from songket patterns. Songket motifs are the main elements that decorate songket pattern. The beauty of songket lies in the elaborate design of the patterns and combination of motifs that are intricately woven on the cloth. After experimenting on thirty songket pattern images, the proposed method achieved promising extraction of the songket motifs.