The process to divide or partition a color image into a set of non- overlapping regions (segments) is called color image segmentation. Color image segmentation can be performed through clustering process by con- sidering the features of each pixel as a pattern and a set of pixels, having similar features or characteristics as a cluster ( segment). Generally, the effectiveness of a clustering algorithm depends on the number of clusters (should be known in advance), initialization of the search space and the searching behaviour of the algorithm. In this work, clustering based algorithms are proposed for color image segmentation which not only determine the number of clusters automat- ically, but also generate compact and well separated segments. First, a hybrid genetic algorithm, called Spatial Fuzzy Genetic Algorithm (SFGA) is proposed which incorporate the colour and spatial information to optimize the fuzzy separation and global compactness simultaneously. The Self Organizing Map (SOM) is adopted to find out the number of clusters (segments) automatically. To initialize the SOM network and SFGA to the productive regions, the dominant peaks in the color his- togram of the wavelet transform image are determined. The problem of over-segmentation is handled with a simple pruning technique. The second contribution is the incorporation of objective function i.e. the ratio of multiple cluster’s overlap to the fuzzy separation into genetic algorithm called Dynamic Genetic Algorithm (DGA). DGA is capable to adjust the number of clusters automatically. Finally, the segmenta- tion of color images are performed by Modified Adaptive Differential Evolution Algorithm (MoADE). MoADE has the ability to automat- ically adjust the crossover and mutation parameters according to the underlying distribution. Moreover to reduce the computational cost the MoADE is applied to the superpixel segmented image. An opposition based strategy is adopted to initialize the population to the productive areas in the search space. The effectiveness of the proposed approaches are tested on Berkeley Im- age Segmentation Database and Benchmark (BSD) with comprehen- sive quantitative and qualitative evaluations. The experimental results demonstrate that the proposed image segmentation methods perform better when applied to complex color images.
"یہ راستہ آگے چل کر چشمے کی طرف نکلتا ہے " راہگیر یہ کہ کر آگے بڑھ گیا. میں شش و پنج میں مبتلا اسے جاتے ہوئے دیکھتا رہا پھر آہستہ آہستہ رستے پر گامزن ہو گیا. اس نے کہا تھا کہ. تھوڑے فاصلے پر چشمہ ہے اور چشمے کا نام و نشان دکھائی نہیں دیتا تھا. میں تھکن, پژ مردگی اور مایوسی سے پہلے ہر حال میں چشمے تک پہنچنا چاہتا تھا. پختہ سڑک کے بعد یہ راستہ زائرین کو پیدل طے کرنا پڑتا تھا. آخر کار چشمہ آ گیا . میں نے پانی پیا اور وہیں زمین پر بیٹھ گیا. اُس نے مجھے یہاں تک آنے کے لیے کہا تھا, اس کے بعد مزار تک پہنچنے کا فاصلہ مجھے اس کے ساتھ طے کرنا تھا. وہ میرے بعد چشمے پر پہنچا اور آتے ہی بولا " آؤ چلیں " ہم چل پڑے اور راستے میں باتیں کرتے رہے. ایک طویل فاصلہ طے کرنے کے بعد اس نے دور سے مزار کی طرف اشارہ کیا اور پھر نا معلوم منزل کی جانب گامزن ہو گیا۔ میں مزار کی طرف بڑھنے لگا ۔ جلد ہی مجھے احساس ہو گیا۔ میں جتنا مزار کے قریب جاتا ہوں ۔ مزار اتنا ہی مجھ سے دور ہو جاتا ہے ۔ میں نے تیز بھاگنے کی کوشش کی اور میرا سانس پھول گیا. میرے پاؤں بوجھل ہونے لگے اور میں حسرت و یاس کے عالم میں مزار کی طرف دیکھتا رہا. مجھے معلوم ہو گیا کہ مجھے باریابی کی اجازت نہیں ہے. میں اپنے بعد آنے والوں کو مزار کی سمت جاتے دیکھتا ہوں۔ سب مجھے حیرت سے دیکھ کر گزر جاتے ہیں۔ میں خستہ حال اور تہی دست ایک عمر سے وہاں پر رکا ہوا ہوں.
Religious Institutions in Pakistan play a vital role in the education of Arabic Language, its dissemination and molding of the religious ethos. Being important centers for the proliferation of the Arabic Language these institutions have specific system for the teaching of Arabic language and it’s methodology. The aim of the paper was to investigate the teaching methods established by Wifaq-ul-Madaris Al- Arabia, Multan Pakistan for the teaching of Arabic literature to the students of Religious Institutions, working under its umbrella. The paper found two major aspects; firstly, to analyze the contents of rabic poetry in various sessions; Secondly, to study various prevailing teaching methods of Arabic poetry in these religious institutions. The paper also explored its solid and fragile features from both aspects and recommends suggestions for the promotions of imparting the educations of Arabic Poetry in religious institutions.
This thesis discusses the electrical response of submicron GaAs MESFETs and HEMTs to develop a physical model. Nine different FET models have been presented and their ability to simulate submicron GaAs MESFET characteristics are checked. To demonstrate the validity of a model, I-V characteristics of short channel MESFETs are simulated and compared with experimental data. The accuracy of a model is reported by evaluating its RMS error values. A comprehensive new model is developed to simulate I-V characteristics of short channel GaAa FETs. It has been demonstrated that the proposed model is a comprehensive one, capable of simulating DC characteristics of GaAs MESFETs including those having significant non-ideal Schottky barrier response. The model has also been applied successfully to I-V characteristics of GaAs HEMTs. The Schottky barrier interfacial layer dependent performance of submicron GaAs MESFETs has been discussed by using their output and transfer characteristics. The mobility of carriers, scattering from the channel into the Schottky barrier gate, increases significantly for the devices which have a relatively thicker interfacial layer. The negative effects of increased carriers’ mobility from MESFET Schottky barrier gate are discussed and a plausible explanation is given for reduced barrier lowering in the presence of interfacial layer. Based on the proposed explanation the definition of threshold voltage has been redefined involving the concept of interfacial layer thickness. A technique is developed to estimate intrinsic small signal parameters of GaAs MESFETs and HEMTs. In the proposed technique DC characteristics are first evaluated. Once a good DC match is attained then small signal parameters are evaluated. To check the validity of the proposed technique submicron GaAs MESFETs and HEMTs of varying gate length have been simulated. It has been shown that the proposed method is accurate as well as efficient in estimating AC parameters of GaAs FETs by using their DC characteristics, and could be employed as a useful tool in device simulation software.