Haniza Yazid

Gradient based adaptive thresholding

For images with poor and non-uniform illumination, adaptive thresholding is required to separate the objects of interest from the background. In this paper a new approach to create an adaptive threshold surface is proposed to segment an image. The technique is inspired by the Yanowitzs method and is improved upon by the introduction of a simpler and more accurate threshold surface. The method is tested on several images of different patterns with varying illumination and the results are compared to the ones produced by a number of adaptive thresholding algorithms. In order to demonstrate the effectiveness, the proposed method had been implemented in medical and document images. The proposed method compares favorably against those using watershed and morphology in medical image and favorably against variable threshold and adaptive Otsus N-thresholding for document image.

Illumination normalization of non-uniform images based on double mean filtering

In segmentation process, non-uniform illumination problem can affect the segmentation result. In this paper, a new method is proposed to solve the problem based on double mean filtering. By applying a combination between mean and threshold value, the varying background is normalized. This proposed method had been experimented with a few badly illuminated images and the result is evaluated by using Misclassification Error (ME), Sensitivity and Specificity. Based on the ME results, proposed method increases the segmentation correction to 88.27%. Besides that, the sensitivity and specificity of proposed method obtained is 94.56190% and 98.57924% and for classical Otsu is 90.30550% and 61.85435%.

A comparison of circular object detection using Hough transform and chord intersection

In this paper, the circular Hough transform (CHT) and the chord intersection have been used to find the circular object in the feature extraction process. The chord intersection technique does not require any gradient information which may be sensitive to noise meanwhile for the CHT technique, the gradient information has been used. In this research, the coconut was selected as the object of interest. 40 images have been experimented to evaluate the performance of the techniques and the detection rate for the CHT is 92.5% and 85% for the chord intersection technique. The average computational time for chord intersection technique is 0.1495s by CPU (AMD Athlon 64x2 Dual core 3800) 2GHz meanwhile CHT consumed more time, 2.3871s in detecting the circular pattern.

Conversion of the Retinal Image Using Gray World Technique

Retinal images are routinely acquired and assessed to provide diagnostic for many important diseases like diabetic retinopathy. People with proliferative retinopathy can reduce their risk of blindness by 95 percent with timely treatment and appropriate follow-up care. The color constancy is used in this context to define the ability of the visual system to estimate an object color transmitting an unpredictable spectrum to the eyes. In this paper, a Gray World method was proposed by assuming the average of the surface reflectance of a typical scene is some pre-specified value. The main idea based on illumination estimated using the statistical region data. The effectiveness of the Gray Word method and normal gray technique was calculated by using Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR). The Gray World achieved the highest PSNR and lowest MSE proved that the image quality was improved. The proposed method can be used to help the ophthalmologist to detect a lesion in the retinal image automatically. Through the contrast variation in retinal images, the disease can be recognized very well.

A review: Comparison between different type of filtering methods on the contrast variation retinal images

Retinal images are obtained using a fundus camera in order to evaluate many important diseases such as Diabetic Retinopathy and Glaucoma. Sometimes, the images appear as uniformly illuminated, have luminosity and contrast variability. This problem can be severely compromised the diagnostic process and the results, especially if automated computer-based procedure is used to derive the parameters of diagnostic. Many researchers propose different approaches to normalize the badly illuminated images based on filtering techniques. In this paper, we compare a six (6) type of filtering techniques and applied to the retinal images from Digital Retinal Images for Vessel Extraction (DRIVE) database to adjust the contrast variation and illumination in order to produce a better diagnostic result. The result performance is evaluate based on Signal Noise Ratio (SNR) and Mean Square Error (MSE) is compared to the other filtering methods. From the result, the Homomorphic filtering based on high pass filter obtained higher SNR value which is 3.093 and the lowest in MSE which is 71267.51.

Luminosity Correction Using Statistical Features on Retinal Images

Retinal fundus image is important for the ophthalmologist to identify and detect many vision-related diseases, such as diabetes and hypertension. From an acquisition process, retinal images often have low gray level contrast and low dynamic range. This problem may seriously affect the diagnostic process and its outcome, especially if an automatic computer-based procedure is used to derive diagnostic parameters. In this paper, a new proposed method based on statistical information such as mean and standard deviation was studied. The combination of local and global technique was successful to detect the luminosity region. Then, a simple correction intensity equation was proposed in order to replace the problem intensity. The results of the numerical simulation (SNR = 2.347 and GCF = 4.581) indicate that the proposed method effective to enhance the luminosity region. Implications of the results and future research directions are also presented. Keywords: Detection, Luminosity, Retinal, Statistical.

Combination of Gray-Level and Moment Invariant for Automatic Blood Vessel Detection on Retinal Image

Segmentation of blood vessels in the retinal is a crucial step in the diagnosis of eye diseases such as diabetic retinopathy and glaucoma. This paper presents a supervised method for automatic segmentation of blood vessels in retinal images. The proposed method based on a hybrid combination between Gray-Level and Moment Invariant techniques. There are four steps involved, whereas preprocessing, feature extraction, classification, and post-processing. In the preprocessing, three stages are performed include vessel central light reflex removal, background homogenization, and vessel enhancement. The 7-D vector feature extraction was performed to compute that compose of gray-level and moment invariants-based features for pixel representation. The decision tree is used for classification step that characterized the pixel based on vessels and non-vessels. The final step is the post-processing which will remove the small artifacts appears after classification process. The proposed method was compared to the Vascular Tree method and Morphological method. Based on the objective evaluation, the proposed method achieved (sensitivity = 98.589, specificity = 55.544 and accuracy = 96.197).

Dual tree complex Wavelet Packet Transform based infant cry classification

A new method has been implemented based on Dual Tree Complex Wavelet Packet Transform (DT-CWPT) feature extraction for infant cry signal classification. The infant cry signals were decomposed into five levels using DT-CWPT. A total of 124 energy features and 124 Shannon entropy features were extracted from each sub-band. Two classifiers Extreme Learning Machine (ELM) and Support Vector Machine (SVM) were used to classify the infant cry signal based on the extracted features. Three category of two-class experiments were conducted in this paper (asphyxia versus normal, hunger versus pain, and deaf versus normal). The results demonstrate that the DT-CWPT feature extraction and classification methods give a high accuracy of 97.87%, 87.26%, 100.00% for asphyxia versus normal, hunger versus pain, and deaf versus normal respectively.

A Novel Contrast Enhancement Technique Based on Combination of Local and Global Statistical Data on Malaria Images

Malaria appears to be one of the main reasons for detrimental health issue at the global scale that is responsible for approximately half a million deaths every year. As the cases of malaria seem to escalate at an annual rate, it is vital to provide a rapid and accurate diagnosis through manual microscopic assessment in the attempt to control the spread of malaria. Nevertheless, varied staining steps and noise disruptions can cause inaccurate diagnosis due to wrong interpretation. Hence, to address such issues, this study investigated the performance upon removing background noise and the method of correcting illumination that has an impact upon segmentation for a computer-assisted diagnostic system. The findings display that the technique of based on Otsu threshold and statistic data used to enhance the contrast image as to determine cells infected by the malaria parasite, in comparison to other methods. In fact, this method was tested on 450 malaria images, which consisted of P. Vivax, P. Falciparum, and P. Knowlesi species at the stages of trophozoite, schizont, and gametocyte. As a result, the HSE approach yielded 1.31 for Global Contrast Factor (GCF), while 10.56 for Signal Noise Ratio (SNR).

Improved binary dragonfly optimization algorithm and wavelet packet based non-linear features for infant cry classification

BACKGROUND AND OBJECTIVE Infant cry signal carries several levels of information about the reason for crying (hunger, pain, sleepiness and discomfort) or the pathological status (asphyxia, deaf, jaundice, premature condition and autism, etc.) of an infant and therefore suited for early diagnosis. In this work, combination of wavelet packet based features and Improved Binary Dragonfly Optimization based feature selection method was proposed to classify the different types of infant cry signals. METHODS Cry signals from 2 different databases were utilized. First database contains 507 cry samples of normal (N), 340 cry samples of asphyxia (A), 879 cry samples of deaf (D), 350 cry samples of hungry (H) and 192 cry samples of pain (P). Second database contains 513 cry samples of jaundice (J), 531 samples of premature (Prem) and 45 samples of normal (N). Wavelet packet transform based energy and non-linear entropies (496 features), Linear Predictive Coding (LPC) based cepstral features (56 features), Mel-frequency Cepstral Coefficients (MFCCs) were extracted (16 features). The combined feature set consists of 568 features. To overcome the curse of dimensionality issue, improved binary dragonfly optimization algorithm (IBDFO) was proposed to select the most salient attributes or features. Finally, Extreme Learning Machine (ELM) kernel classifier was used to classify the different types of infant cry signals using all the features and highly informative features as well. RESULTS Several experiments of two-class and multi-class classification of cry signals were conducted. In binary or two-class experiments, maximum accuracy of 90.18% for H Vs P, 100% for A Vs N, 100% for D Vs N and 97.61% J Vs Prem was achieved using the features selected (only 204 features out of 568) by IBDFO. For the classification of multiple cry signals (multi-class problem), the selected features could differentiate between three classes (N, A & D) with the accuracy of 100% and seven classes with the accuracy of 97.62%. CONCLUSION The experimental results indicated that the proposed combination of feature extraction and selection method offers suitable classification accuracy and may be employed to detect the subtle changes in the cry signals.