I. INTRODUCTION Data Mining has become moreimportant in society due to the large amount of data and changing such datainto the useful information and knowledge. Extracting or mining knowledge fromhuge collection of data is called Data mining. The main goal of data mining isto mine information from the set of data and translate it into anunderstandable structure for future use. Data mining is one among the knowledgediscovery process.  Knowledge discovery has sequence of steps as:Data cleaning, Data integration, Data selection, Data transformation, DataMining, Pattern evaluation, Knowledge presentation. It uses techniques that areused to extract data patterns. Data mining system has engine which comprises ofset of functions for tasks such as characterization, association andcorrelation analysis, classification, prediction.

Nowadays web mining has gained moreattention of users with its interfaces and large quantity of knowledge on themarket. Extracting patterns that are accessed by the users in distributedinformation environment is called Web mining, Web search based on the singlekeyword may outputs hundreds of web page links containing the keyword, but mostof the links will be weakly related to which the user want to search. Extracting Frequent Patterns leadsto the discovery of interesting associations. Frequent patterns are thepatterns which occur frequently. Market basket analysis is an example offrequent itemset. Association analysis is the method which is used to findinteresting relationship hidden in large amount of data. Association analysisare used to cover relationship among related data in the database, relationaldatabase or other information repository.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

Association rules are used to findthe relationships between the objects which are frequently used together.Applications of association rules are basket data analysis, classification,cross-marketing, clustering, catalog design, and loss-leader analysis etc. Inthis paper we are using item sets as images where related images are displayedbased on content based (image itself ) and image name based. For example, if thecustomer buys rice then he may also buy dhal. If the customer buys mobile thenhe may also buy memory card.

There are two measures that association rulesuses, support and .confidence. It identifies the relationships generated byanalyzing data for frequently used patterns. Association rules are usuallyneeded to satisfy a user-specified minimum support and a user –specified minimumconfidence at the same time.          Nowadayswith the use of various digital cameras and the rapid growth of social mediafor internet-based photo sharing, recent years have witnessed an explosion ofthe number of digital photos captured and stored by users. Major issue that hasto be taken care is the recognition of images that is to identify or verify theimages using the database where the images are stored.

Image recognition is animportant part of the capability of human perception system. The initial workon image recognition can be traced back at least to the 1950s in psychology andto the 1960s in the engineering literature. Some of the earliest studiesinclude work on facial expression of emotions by Darwin.  Later many conceptswere used in the recognition of images such as identification number, race,age, gender, facial expression, or speech may be used in narrowing the search(enhancing recognition).

The solution to the problem involves segmentation offaces (face detection) from cluttered scenes, feature extraction from the faceregions, recognition, verification and also indexing may be applied on images.In identification problems, the input to the system may be given as image orthe name of the image, and the system outputs the similar images from adatabase of known individuals or else outputs null, whereas in verificationproblems, the system needs to confirm or reject the identity of the inputimage.  In most cases photos shared by users on theweb are facial images. Some facial images are label with names, some may beweakly labeled and some are not labeled properly. This motivated to an importanttechnique that is to find facial images automatically. This can be useful tomany applications on web and online photo-sharing sites can automaticallylabels user uploaded photos to provide online photo search A method ispresented for giving label to facial image by mining the web, where a hugenumber of weakly labeled images are available freely in internet. This aims tothe automated face annotation(identification) task by taking the advantage ofcontent-based image retrieval (CBIR) and search based image retrieval (SBIR)techniques in mining the large amount of poorly labeled  images on the internet. This framework ismodel-free and data-driven.

The main motives of these schemes are to assigncorrect name labels to a given image query. For given a novel facial image forannotation, first we have to retrieve a short list of top n most same facialimage pixels from a poorly labeled facial image database, and then annotate thefacial image by the names(labels) associated with the top n facial images ofsame pixel value(binary value).  Onechallenge faced by CBIR and SBIR techniques is how to effectively identify andto short list similar facial images and their weak labels for the face nameannotation task. To solve this, we use a novel updated unsupervised labelrefinement (ULR) scheme by considering machine learning techniques. We alsopropose Cluster based approximation algorithm (CBA) and Association rule basedapproximation (ABA) algorithm to improve the efficiency. We can also providefacility to search similar images by giving input in the form of image. II.

LITERATURE SURVEY Dayong Wang, StevenC.H. Hoi, Ying He, and Jianke Zhu has proposed –  Mining Weakly Labeled Web Facial Images forSearch-Based Face Annotation gives a framework of search-based face annotation(SBFA) by mining weakly labeled facial images that are freely available on theWorld Wide Web (WWW). This mainly exploits the list of most similar facialimages and their labels that are noisy that uses unsupervised label refinement(ULR) approach for refining the labels of web facial images using machinelearning techniques. Zhong Wu, Qifa Key, Jian Suny, Heung-Yeung Shumyhas proposed – Scalable Face Image Retrieval with Identity-BasedQuantization and Multi-Reference Re-ranking, which aims to build a scalable face image retrieval system and develops a newscalable face representation using both local and global features. In theindexing stage, exploits special properties of faces to design newcomponent-based local features,which are subsequently quantized into visualwords using a novel identity-based quantization scheme.  Preeti Chouhan, Mukesh Tiwari has proposed – Feature Extraction Techniques forImage Retrieval Using Data Mining and Image Processing Techniques provides witha basic informatory review on the applied fields of data mining which is variedinto manufacturing, telecommunication, education, fraud detecting and marketingsector.

Includes some of the methods like clustering, correlation, associationand neural network and also provides concepts on Image mining. Image miningdeals with  association of image data andextraction of hidden data.C.

Ganesh, B.Sathiyabhama, T.Geetha has proposed- Fast Frequent Pattern Mining Using Vertical Data Format forKnowledge Discovery, provides Apriori based techniques,Frequent Pattern growth (FP-growth) and Equivalence CLASS Transformation(ECLAT) are the widely used approaches used in extracting frequent patterns.

Also quantitative investigation of changing the format stream is done forbetter result in less computational time. II.METHODOLOGY  A.Existing System Inthe Existing system, object recognition techniques is used to trainclassification models from human-tagged training images or attempt to show thecorrelation between annotated keywords and images. Given limited training data,semi-supervised learning methods have been used for image identification in classicalclassification models. Limitations:1.Similar clear Images were not displayed using local binary system. 2.

Poorlyappeared images or poorly labeled images are difficult to identify. 3. Alwaysproduces approximate results.4. Therewas no ranking (count) scheme.

 B. ProposedSystemThispaper mainly gives a framework for search-based and content based imageretrieval techniques by mining weakly named images that are available . Sincethe images are poorly labeled, it will be difficult to identify the similarimages. So to identify the poorly labeled similar images, we have proposedupdated unsupervised label refinement (ULR) approach . To perform search onimages we are using ULR algorithm having the binary format of the images. Searchcan be done based on the name of image or the image itself, if the match isfound in the unit then the similar images are displayed otherwise the output isnull. Grouping of images are done using cluster approximation. Also count ofthe images are monitored based on user clicks for the respective images whichis searched.

Advantages:1.Similar Clear Images were retrieved based on image itself or the name of theimage. 2. Easyto retrieve the images since the names are given to the images.

3.Produces accurate results.4.

Thereis ranking (count) scheme based on the number of times the user searched forparticular image We have 4important modules in this process:1.      LabelingImages: Imagesare uploaded by giving label(name) to the images. 2.      Content-Based Image Retrieval: In this module, input is given asimage and outputs group of images that are similar to the input image elseoutputs null.

Query by image content (QBIC) method is used.3.      Search based Image Retrieval: Inthis module, input is given as name and outputs group of images that are similarto the input name else outputs null. Query by image name(QBNC) method is used.

4.      RankingScheme: Count ofthe respective images that are searched are recorded. C.Architecture                                                                    Figure 1Figure 1 illustrates the system flow ofthe proposed framework of search-based face annotation, which consists of thefollowing steps:1.Collectionof images, Labeling and Storing2.Detection and Feature Extraction based on the input3.Performing Indexing and Collect the labeled data using the URL technique4.

Faceannotation where similar Images are retrieved using Cluster analysis5. Faceannotation by ranking scheme using association analysisThe first 3 steps areusually conducted before the test phase of a Image identification task, whilethe last two steps are conducted during the test phase of a Imageidentification task, which usually should be done very efficiently. We brieflydescribe each step below.The first step is the data collection offacial images as shown in Figure 1, in which we collect the images by Googlesearch engine.

Given the nature of web images, the images may be noisy, whichdo not always correspond to the correct name and such images are weakly orpoorly labeled facial images. The second step is to  detect and extract the feature of images , weuse the unsupervised face alignment technique proposed in 4. For facialfeature representation, we extract the GIST texture features 5 to representthe extracted faces. The third step is indexing the extracted features of the imagesby applying some efficient high-dimensional indexing technique. So for this, weuse the locality sensitive hashing (LSH) 6 and  unsupervised learning scheme is used toenhance the label quality of the weakly labeled facial images which isimportant in the search process. The first 3 steps are the phasesinvolved in updated ULR algorithm.The fourth step isgrouping of related images (K similar images) using cluster approximationalgorithm. Last step is Faceannotation by counting the user clicks based on user search of particular imageand this is done using association analysis.

 III.ALGORITHMS                                                                                                             A.     Updated ULR algorithmInput: ImageOutput: Similar Images/ NULLBeginCollection of images, Labeling and StoringDetection and Feature Extraction based on the inputPerforming Indexing and Collect the labeled datausing the URL technique            End            B.    ClusterBased Approximation Algorithm:Thenumber of variables in the extracted image feature are a * b. Where a= numberof facial images in the retrieval database. b= number of distinct names. Inthis paper strategy could be applied in two different phases: Imageretrieval based on  1.

One ison “image itself,” which can be used to separate all the ‘a’ facial images intosimilar group2. Secondone is on “image name,” which can be used to separate the ‘b’ names into agroup.Thenbased on the input which is given the similar images of respective cluster aredisplayed.  Inthis paper k-NN clustering technique is used for clustering the images.

The k-NearestNeighbors algorithm (k-NN) is used for classification andregression. In K-NN, the input consists of the k closest sample data. In k-NNclassification, the output is a class member. An object is classified by apopular counting of its neighbor point .

If k = 1, then the object is assignedto the class of that one nearest neighbor. The property value is the object ink-NN regression. This value is the average of the value of its k nearestneighbors.  C.    AssociationBased Approximation Algorithm:            Here, based on the image retrievalthe count is monitored for every user clicks on the images and that count willbe reflected in the clusters for the further use.

 IV. CONCLUSIONInthis paper a search-based image retrieval and content based image retrievaltechniques are used to mine the huge amount of poorly labeled images that arefreely available on the web. It uses a updated ULR algorithm to identify theimages, Cluster approximation method to group the similar images forscalability and Association analysis method used to monitor the number of timesthe particular image is been searched. All these methods improve theperformances and also scalability without degrading the system performance. Futureenhancement can be done on retrieval of images based on time.