Object Recognition Using Modified Normalized Cross Correlation

Proceedings of the 1st International Conference on Emerging Issues in Technology, Engineering and Science

The segmentation process plays a very important role in the process of detecting and recognizing objects in an image. Although many segmentation methods have been developed, there is no method that can give good results and is generally accepted. This study aims to find a better segmentation method in finding patterns or ground-truths of objects in an image, so that these objects can be easily recognized. Based on the results shown by all the methods being compared, several methods were not successful in showing the pattern of objects contained in the sample image. The segmentation method based on color dissimilarity, which is a method that emulates the way humans recognize an object based on visible color differences, shows the best results compared to other methods. This method is a very suitable method to be used in the process of detecting and recognizing objects in an image.

Segmentation is a fundamental process in digital image processing which has found extensive applications in areas such as medical image processing, compression, diagnosis arthritis from joint image, automatic text hand writing analysis, and remote sensing.The clustering methods can be used to segment any image into various clusters based on the similarity criteria like color or texture. In this research we have developed a method to segment color images using K-means clustering algorithm. K-means clustering algorithm divides the image into K clusters based on the similarity between the pixels in that cluster. In this research we have used Euclidean distance formula to define clusters in K-mean clustering. The proposed method has been applied to a variety of images and conclusions have been drawn.

International Journal of Computer Applications, 2016

The recognition of shape is a method aiming the identification of the motives from raw data in order to take a decision depending on the category assigned to this motive. The method of motive detection by cross-correlation is a very efficient method in recognition of shape. Indeed, it permits to localize an object in an image. But, the value of the cross-correlation depends more of the level of gray of the image. In this article one proposes a solution of this inconvenience based on the Fourier Transform where by exploiting its phase for to localize an object while giving its position in the image.

Informatica, 2022

This technique is able to give the same results and is good for images with JPG, PNG and BMP file types. To provide confidence in the performance of this method, a comparison is made with several segmentation methods. By measuring the similarity index using the Jaccard index, Sørensen-Dice index and BF Score, this color dissimilarity-based segmentation method has the best performance. Meanwhile, when measuring the processing time, this method has the second best performance. Thus, it can be concluded that this method is very suitable for detecting objects in an image. Povzetek: Tehnika segmentacije je bila razvita z imitacijo človeškega koncepta prepoznavanja predmeta, in sicer na podlagi barvnih razlik. Barve se štejejo za različne, če imajo vrednost DeltaE najmanj 8, ki jo dobimo iz razlik v vrednostih R, G ali B barv samih ali v parih. Izvajanje te tehnike kaže enake rezultate za datoteke JPG, PNG in BMP in je v veliko pomoč pri postopku identifikacije predmeta.

International Journal of Computer Applications, 2012

The goal of this paper is to analyse and improve the performance of metrics like Coefficient of Correlation (CoC) and Structural Similarity Index (SSIM) for image recognition in real-time environment. The main novelties of the methods are; it can work under uncontrolled environment and no need to store multiple copies of the same image at different orientations. The values of CoC and SSIM get changed if images are rotated or flipped or captured under bad/highly illuminated conditions. To increase the recognition accuracy, the input test image is pre-processed. First, discrete wavelet transform is applied to recognize the image captured under bad illuminated and dull lightning conditions. Second, to make the method rotation invariant, the test image is compared against the stored database image without and with rotations in the horizontal, vertical, diagonal, reverse diagonal and flipped directions. The image recognition performance is evaluated using the Recognition Rate and Rejection Rate. The results indicate that recognition performance of Correlation Coefficient and SSIM gets improved with rotations and discrete wavelet transform. Also it was observed that CoC with proposed modifications yield better results as compared to state of the art enhanced Principal Component Analysis and Enhanced Subspace Linear Discriminant Analysis.

Object recognition refers to finding a given object in an image or video sequence based on physical properties of the object. Object recognition has been a challenging computer vision problem with various real-world applications from a very long time. In this paper we show Object recognition by correlation filtering consists in calculation of cross correlation function between an test image and noisy query image, this work aims to bring a promising perspective for using correlation filter. In order to reduced noise and perform an optimization algorithm .Based on optimization method, correlation filter has been proposed. The proposed filter has been tested to recognize an object at different noise condition. The proposed filter has been evaluated in terms of peak to side lobe ratio (PSR) and peak to correlation energy (PCE) between the query image and test image for various noise conditions

International Journal of Computer Applications, 2012

Never before in history has image data been generated at such high volumes as it is today. If images are analyzed properly, they can reveal useful information to the users. Image mining deals with the extraction of implicit knowledge, image data relationship, or other patterns not explicitly stored in the images. Image clustering involves the extraction of features from image databases and then application of data mining algorithm to group images. In this paper a data mining approach to cluster the images using color and texture features are proposed. Three techniques are proposed to extract Color feature, using Color Moments, Block Truncation Coding algorithm and histogram method. To extract texture feature concept of Gray Level Co-occurrence Matrix is extended and applied to color images. K-means clustering algorithm is applied to groups the images.

KSII Transactions on Internet and Information Systems, 2011

Content Based Image Retrieval (CBIR) is an approach for retrieving similar images from an image database based on automatically-derived image features. The quality of a retrieval system depends on the features used to describe image content. In this paper, we propose an image clustering system that takes a database of images as input and clusters them using kmeans clustering algorithm taking into consideration color, texture and shape features. Experimental results show that the combination of the three features brings about higher values of accuracy and precision.

Image segmentation divides an image into several constituent components such as color, structure, shape, and texture. It forms a major research topic for many image processing researchers as the applications are endless. Its applications include image enhancement, object detection, image retrieval, image compression, and medical image processing to name a few. The segmentation of color images is necessary for efficient pattern recognition and feature extraction involving various color spaces such as RGB, HSV and CIE L*A*B*, etc. This paper describes the different cluster based segmentation techniques used for segmenting the different color images and the resultant is analyzed with subjective and objective measures. Initially, registered color images are considered as input. Then the cluster based segmentation techniques namely K-Means clustering, Pillar-Kmeans clustering and Fuzzy C-means (FCM) clustering techniques are applied. Further, the segmented image is analyzed with measures such as compactness and execution time. From the experimental results, it has been observed that K-means and Pillar-Kmeans are the most suitable techniques for RGB, HSV and LAB color spaces than the FCM technique.

UNITECH, 2017

In this study, we aimed to detect and count similar objects in an image. Object images are obtained from conditioned environment that camera position and lighting are fully controlled. Variously shaped objects can be in any position and direction on black plane. Model object is selected by user from image on screen. User simply draws a selection box around desired object image. This object image is used for creating template. Gaussian Filter, Pyramid Down and Pyramid Up filters are used for smoothing obtained image. Binary Threshold is using for determine objects boundaries. Template Matching is used for identifying objects. Template image is rotated by certain intervals for detecting same object in any orientation. Cross correlation is evaluated between template and image for detecting object. Found objects are indicated with red box and then count. Algorithm was tested with various bolt, nut and washer images. Detection accuracy is calculated as 92% for test image