Abstract: This paper discusses a new color model for digital image, which can be used to separate low, and high frequencies in the image without loosing any information from the image . The proposed model was used as an initial step to imagesegmentation and the practical results shows the efficiency of using this method to get the actual objects of the color image. The proposed model can be used in different application such as separating low and high frequencies from the image, which are proposed in , and it also can be used to segment different types of color images.
The task of imagesegmentation is to divide an image into a number of non-overlapping regions, which have same characteristics such as gray level, color, tone, texture, etc. Famous techniques of imagesegmentation which are still being used by the researchers are Edge Detection, Threshold, Histogram, Region based methods, and Watershed Transformation. Since images are divided into two types on the basis of their color, i.e. gray scale and color images. Therefore imagesegmentation for color images is totally different from gray scale images, e.g., content based image retrieval, . Also which algorithm is robust and works well is depends on the type of image .
A very important part of the image processing area is imagesegmentation. This refers to the task of partitioning a given image into multiple regions and is typically used to locate and mark objects and boundaries in input scenes. After segmentation the image represents a set of data far more suitable for further algorithmic processing and decision making. Imagesegmentation algorithms are a very broad eld and they have received signi cant amount of research interest A good example of an area, in which image processing plays a constantly growing role, is the eld of medical solutions. The expectations and demands that are presented in this branch of science are very high and dif cult to meet for the applied technology. The problems are challenging and the potential bene ts are signi cant and clearly visible. For over thirty years image processing has been applied to different problems and questions in medicine and the practitioners have exploited the rich possibilities that it offered. As a result, the eld of medicine has seen signi cant improvements in the interpretation of examined medical data. Clearly, the medical knowledge has also evolved signi cantly over these years, as well as the medical equipment that serves doctors and researchers. Also the common computer hardware, which is present at homes, of ces and laboratories, is constantly evolving and changing. All of these factors have sculptured the shape of modern image processing techniques and established in which ways it is currently used and developed. Modern medical image processing is centered around 3D images with high spatial and temporal resolution, which can bring a tremendous amount of data for medical practitioners. Processing of such large sets of data is not an easy task, requiring high computational power. Furthermore, in present times the computational power is not as easily available as in recent years, as the growth of possibilities of a single processing unit is very limited - a trend towards multi-unit processing and parallelization of the workload is clearly visible. Therefore, in order to continue the development of more complex and more advanced image processing techniques, a new direction is necessary.
The Quality of the food materials are identified using the computer aided system. The key idea of the proposed method is to identify the defected parts of food materials. The proposed framework provides the comparison of various filters and the hybrid median filter was selected as the filter with the high PSNR values and was used in the preprocessing stage. Imagesegmentation process namely Colour based binary Imagesegmentation and Particle swarm optimization techniques were compared then their parameters such as accuracy, specificity, sensitivity were measured and found that the color based binary imagesegmentation were well suited for imagesegmentation. Finally the defected parts were being segmented from the original image. The future work is based on feature classification and feature selection of food products by Artificial Neural Networks to classify the quality of food as accepted, rejected and premium.
Machine vision technology in agricultural research started relatively late to achieve high efficiency. The high- precision segmentation algorithm is a hot research problem to realize automatic identification. Our study aimed to find a suitable method of rice planthopper segmentation to eliminate the irrelevant information in images and enhance the detection of relevant information and maximally simplify data. In the statistical model based on object segmentation, an object is assumed to be a realization of a random process, which is governed by the model parameters. Object features obtained from these methods, especially Markov random field (MRF), have been proven to offer a powerful framework for imagesegmentation. MRFs were been introduced by Besag  via their pseudo-likelihood approximation. The Gibbs distribution inherent to MRFs was then exploited. Since this pioneer work, MRFs have been actively used in the image-processing community for modeling spatial correlations. Examples of applications include the segmentation of SAR or brain magnetic resonance images , . Other interesting works involving MRFs for segmentation and classification have been published , , .
Clustering is one of the widely used imagesegmentation techniques which classify patterns in such a way that samples of the same group are more similar to one another than samples belonging to different groups [3,9,10]. There has been considerable interest recently in the use of fuzzy clustering methods, which retain more information from the original image than hard clustering methods. Fuzzy C-means algorithm is widely preferred because of its additional flexibility which allows pixels to belong to multiple classes with varying degrees of membership. But the major operational complaint is that the FCM technique is time consuming.The drawback of the FCM is improved by the improved FCM algorithm.
Determination of the optimal number of cluster in clustering process is also a challenge. There are many indexes to determine the cluster validity such as Dunn Index, Silhouette Index, and others. Validity Measure has been widely applied as cluster validity measurement in imagesegmentation application ,. In this study, the cluster validity measurement will be done in 2 ways, by Validity Measure (VM) and Davies-Bouldin Index (DBI). The validity measurement is done by forming 2 clusters to 10 clusters, then calculate the validity of each cluster. Cluster with the minimum values of DBI or VM indicates the cluster is well separated , it means that number of cluster is the most optimal cluster.
The segmentation of images into homogeneous regions is an important area of research in computer vision. Image thresholding, which is a popular technique for imagesegmentation, is also regarded as an analytic image representation method . This technique plays an important role in many of the tasks that are required for pattern recognition, computer vision, and video retrieval . Image thresholding is computationally simpler than other existing algorithms, such as boundary detection or region dependent techniques [3-6]. Its aim is to find an appropriate threshold for separating the object of interest from the background. The output of a thresholding process is a binary image in which all of the pixels with gray levels higher than the determined threshold are classified as object and the remaining of pixels are assigned to
All parameters analyzed determine the best choice for ANN configuration have following organization: 3 inputs RGB, 3 neurons in the hidden layer, RL of 0.01 and CM of 0.9. The comparison of the ANN with other system with two ANN, one with filter function and other with segmentation, demonstrated that results achieved with just one ANN (optimized) is better than with two ANN for treatment of impulsive noise (Salt and Pepper). ANN optimized obtained better results than with linear filters by average simple and weighted. ANN with few neurons can obtain results satisfactory for imagesegmentation in the recognition of nutrients deficiency.
Abstract— This paper proposes a new segmentation approach based on graph cuts. However, we take advantage of it in a different way using photometric (color and texton) and geometric information to analyze and segment images. These additional information provide better knowledge of the image’s internal structure. Hence, they lead to a significant improvement in segmentation quality, while require fewer instruction from user interactions. Some experimental results are presented in the variety contexts of imagesegmentation giving high accurate segmentation comparing with other methods.
In this work, an algorithm was presented in order to obtain better results in cell imagesegmentation, with the assistance of a human user. Even though a semi-automatic is not the perfect situation, it makes the application to be more tolerant to any kind of situation. On one hand, it allows to obtain results at a faster rate than a totally manual segmentation. On the other hand, it has the potential to give better results than purely automatic applications, since the user can alter the final result, adjusting the contour to the correct position with a quick and simple procedure.
Abstract —Data clustering techniques are often used to segment the real world images. Unsupervised imagesegmentation algorithms that are based on the clustering suffer from random initialization. There is a need for efficient and effective imagesegmentation algorithm, which can be used in the computer vision, object recognition, image recognition, or compression. To address these problems, the authors present a density-based initialization scheme to segment the color images. In the kernel density based clustering technique, the data sample is mapped to a high-dimensional space for the effective data classification. The Gaussian kernel is used for the density estimation and for the mapping of sample image into a high- dimensional color space. The proposed initialization scheme for the k-means clustering algorithm can homogenously segment an image into the regions of interest with the capability of avoiding the dead centre and the trapped centre by local minima phenomena. The performance of the experimental result indicates that the proposed approach is more effective, compared to the other existing clustering-based imagesegmentation algorithms. In the proposed approach, the Berkeley image database has been used for the comparison analysis with the recent clustering-based imagesegmentation algorithms like k- means++, k-medoids and k-mode.
The neural network has not been used for the first time to segment an image. Previously imagesegmentation has been done using the MLSONN architecture. Many other algorithms are there for this purpose. The MLSONN architecture has some draw backs. As in this architecture in the back propagation part there is a recurrent loop connecting the output layer to the input layer which basically increase the complexity. This project work has been done based on the research that find the new BDSONN architecture in which the output is feed backed to hidden layer instead of input layer. It reduces so many computation burdens as much possible. That is the main advantage of this architecture that has been implemented in this project using the language Visual Basic 6.In the coming future BDSONN architecture will be widely used to perform imagesegmentation in the world. So the goal of our project is to put a look on this research and try to visualize our thought in this concern and this resource should be helpful.
according to the result of a heat diffusion process in which the seeded pixels are considered to be the heat sources and the heat diffuses on the image starting from the sources. In 2012, ChuanLong Li et al.  propose a novel fuzzy c-means imagesegmentation algorithm. Its effectiveness is due to two mechanisms. The first mechanism is the replacement of the Euclidean distance traditionally used to measure similarity of the image pixels by a novel similarity measure which is considered spatial neighbourhoods using Gaussian kernel, and thus their method becomes less sensitive to the noise of the image. The second mechanism is not requirement of any similarity penalty term in FCM's objective function as some FCM's variants to reduce the influence of noise on the result of imagesegmentation. In 2009, Yi-hua Lan et al.  proposed a novel imagesegmentation method based on random walk model. First of all, they down- sampling the original large image to the small image which can be solved fast, then the small imagesegmentation leads to sparse linear equations of much smaller scale. After getting the solution, the probability results will be up-sampling to the up layer, and then solve the sparse linear equations in this layer; repeating this up-sampling process until to the top layer which is the original image. At last, segment the final probability image with a pre-set threshold. In 2012, Forsthoefel, D. et al.  describe a new, highly efficient imagesegmentation technique called leap segmentation that builds a new image representation where individual pixel data is replaced with a map of chromatic- and illumination-similar regions that are adjacent but not necessarily contiguous. In 2011, Shaohua Zhu et al.  suggest that due to color image contain more information than grayscale image, color image processing is being paid more and more attention. The methods of grayscale imagesegmentation just consider the brightness and neglect hue, saturation and other important information. In 2008, Hasanzadeh, M. et al.  proposed a new segmentation method based on a combination of fuzzy connectedness and fuzzy clustering called membership connectedness, by which the spatial relation of image pixels is constructed in the related membership domain. They have proposed a new fuzzy connectedness relation for imagesegmentation in membership domain which outperforms the previously defined relation in noisy images.
Imagesegmentation is the most important step and a key technology in image processing, and it will directly affect subsequent processing. With mathematical theories, imagesegmentation has achieved great progress and a lot of novel segmenting algorithms have been proposed. But most algorithms have their own drawbacks. As for cell images, owing to the complex nature, it still remains a challenging task to segment and count them. Extensive researches show that PCNN is very suitable for imagesegmentation and can get a perfect result, which is primarily due to the inherent ability of PCNN.
Abstract - In image processing and image analysis the final result is obtained by combining information from various resources. To obtain better result from combined information, image registration plays an important role, which is a earlier step in imagesegmentation. Generally it follows feature extraction, feature matching, transformation and sampling, dominant extracted features and matching algorithm gives the better registration accuracy. In proposed system contourlet transform and mutual information is combined to increase the accuracy in registration process. Contourlet transform extracts efficient curves and edges from MRI images, these features from MRI brain images helps to match the information using mutual information. Performance comparison of proposed results shows high performance using Contourlet transform.
From the last decade, the multi-scale imagesegmentation is getting a particular interest and practically being used for object-based image analysis. In this study, we have addressed the issues on multi-scale imagesegmentation, especially, in improving the performances for validity of merging and variety of derived region’s shape. Firstly, we have introduced constraints on the application of spectral criterion which could suppress excessive merging between dissimilar regions. Secondly, we have extended the evaluation for smoothness criterion by modifying the definition on the extent of the object, which was brought for controlling the shape’s diversity. Thirdly, we have developed new shape criterion called aspect ratio. This criterion helps to improve the reproducibility on the shape of object to be matched to the actual objectives of interest. This criterion provides constraint on the aspect ratio in the bounding box of object by keeping properties controlled with conventional shape criteria. These improvements and extensions lead to more accurate, flexible, and diverse segmentation results according to the shape characteristics of the target of interest. Furthermore, we also investigated a technique for quantitative and automatic parameterization in multi-scale imagesegmentation. This approach is achieved by comparing segmentation result with training area specified in advance by considering the maximization of the average area in derived objects or satisfying the evaluation index called F-measure. Thus, it has been possible to automate the parameterization that suited the objectives especially in the view point of shape’s reproducibility.
Medical imagesegmentation is still an on-going research topic. The wide range of imaging pro- tocols with their respective scanning parameters makes it difficult to have an unique solution for imagesegmentation [1, 2]. Moreover, the performance of segmentation methods is also impaired by the presence of pathologies. For example, MR images acquired with sequences such as DIXON or IDEAL [3, 4] produces two images, fat and water, which are used to study fat infiltration in the musculoskeletal system. However, the low contrast quality of the edges that describe the interfaces between muscles affects the performance of the segmentations algorithms.
E XTRACTING the boundary and register the given im- ages are the most important tasks in image processing, image analysis, and computer vision. There have been nu- merous techniques developed to solve imagesegmentation and registration problems. In the past, solutions of these two problems have been studied separately from each other.
Abstract —Fingerprint imagesegmentation is one of the important preprocessing steps in Automatic Fingerprint Identification Systems (AFIS). Segmentation separates image background from image foreground, removing unnecessary information from the image. This paper proposes a new fingerprint segmentation method using Haar wavelet and Kohonen’s Self Organizing Map (SOM). Fingerprint image was decomposed using 2D Haar wavelet in two levels. To generate features vectors, the decomposed image was divided into nonoverlapping blocks of 2x2 pixels and converted into four elements vectors. These vectors were then fed into SOM network that grouped them into foreground and background clusters. Finally, blocks in the background area were removed based on indexes of blocks in the background cluster. From the research that has been carried out, we conclude that the proposed method is effective to segment background from fingerprint images.