Leena Lepistö

Multiscale Fourier descriptors for defect image retrieval

Abstract Shape is an essential visual feature of an image and it is widely used to describe image content in image classification and retrieval. In this paper, two new Fourier-based approaches for contour-based shape description are presented. These approaches present Fourier descriptors in multiple scales, which improves the shape classification and retrieval accuracy. The proposed methods outperform ordinary Fourier descriptors in the retrieval of complicated shapes without increasing computational cost.

Multiscale Fourier descriptor for shape-based image retrieval

The shapes occurring in the images are important in the content-based image retrieval. We introduce a new Fourier-based descriptor for the characterization of the shapes for retrieval purposes. This descriptor combines the benefits of the wavelet transform and Fourier transform. This way the Fourier descriptors can be presented in multiple scales, which improves the shape retrieval accuracy of the commonly used Fourier-descriptors. The multiscale Fourier descriptor is formed by applying the complex wavelet transforms to the boundary function of an object extracted from an image. After that, the Fourier transform is applied to the wavelet coefficients in multiple scales. This way the multiscale shape representation can be expressed in a rotation invariant form. The retrieval efficiency of this multiscale Fourier descriptor is compared to an ordinary Fourier descriptor and CSS-shape representation.

Multiscale Fourier descriptor for shape classification

The description of object shape is an important characteristic of an image. In image processing and pattern recognition, several different shape descriptors are used. In human visual perception, shapes are processed in multiple resolutions. Therefore, multiscale shape representation is essential in shape based image classification and retrieval. In the description of an object shape, the multiresolution representation provides also additional accuracy to the shape classification. We introduce a new descriptor for shape classification. This descriptor is called the multiscale Fourier descriptor, and it combines the benefits of a Fourier descriptor and multiscale shape representation. This descriptor is formed by applying a Fourier transform to the coefficients of the wavelet transform of the object boundary. In this way, the Fourier descriptor can be presented in multiple resolutions. We performed classification experiments using three image databases. The classification results of our method are compared to those of Fourier descriptors.

Rock image classification using color features in Gabor space

In image classification, the common texture-based meth- ods are based on image gray levels. However, the use of color information improves the classification accuracy of the colored tex- tures. In this paper, we extract texture features from the natural rock images that are used in bedrock investigations. A Gaussian band- pass filtering is applied to the color channels of the images in RGB and HSI color spaces using different scales. The obtained feature vectors are low dimensional, which make the methods computation- ally effective. The results show that using combinations of different color channels, the classification accuracy can be significantly

Classification method for colored natural textures using Gabor filtering

The common methods of texture analysis are based on the gray levels of the texture image. However, the use of color information improves the classification accuracy of colored textures. In the classification of non-homogenous natural textures, human perception of texture and color are important. Therefore, the color space and texture analysis method should be selected to correspond to human vision. We present an effective method for the classification of colored natural textures. The natural textures are often non-homogeneous and directional, which makes them difficult to classify. In our method, multiresolution Gabor filtering is applied to the color components of the texture image in HSI color space. Using this method, the colored texture images can be classified in multiple scales and orientations. Experimental results show that the use of color information improves the classification of natural textures.

IMAGE CORRELOGRAM IN IMAGE DATABASE INDEXING AND RETRIEVAL

In this paper we present a new approach to image indexing and retrieval based on image correlogram. We show that in comparison with usually used approaches, i.e. image histogram and autocorrelogram, image correlogram gives significantly better results in image retrieval. These results are achievable without increasing computational cost in image indexing or retrieval. In the experimental part of this paper the retrieval performance of image correlogram is compared to that of image autocorrelogram and image histogram.

Fourier-Based Object Description in Defect Image Retrieval

Image retrieval has nowadays several industrial applications. In these imaging applications, which typically use large image archives, the matter of computational efficiency is essential. Therefore, compact and efficient features are required to describe the visual content of the images. In this paper, we introduce a new Fourier-based object descriptor that combines the shape and color information of the objects occurring in images into a single low-dimensional descriptor. The experiments performed with an industrial surface defect image database show that the proposed descriptor is an accurate and computationally light approach to object description in a real-world retrieval problem.

Efficient Fourier shape descriptor for industrial defect images using wavelets

The use of image retrieval and classification has several applications in industrial imaging systems, which typically use large image archives. In these applications, the matter of computational efficiency is essential and therefore compact visual descriptors are necessary to describe image content. A novel approach to contour-based shape description using wavelet transform combined with Fourier transform is presented. The proposed method outperforms ordinary Fourier descriptors in the retrieval of complicated industrial shapes without increasing descriptor dimensionality.

Classification of nonhomogenous texture images by combining classifiers

Most of the natural textures are nonhomogenous. In nonhomogenous texture images, the textural features may have strong variations. These variations cause errors in the classification of these images. In this paper we present a novel method for classification of the nonhomogenous textures. The classification method is based on the combination of separate classifiers. The outputs of the separate classifiers are collected into a classification result vector (CRV). This vector is used in the final classification of the texture samples. Using this method, the classification errors caused by variations of feature values can be minimized. The method is tested using nonhomogenous rock texture images. The results show that our method is suitable for classifying nonhomogenous texture samples. It also gives better classification results than the commonly used methods for combining classifiers.

Color-Based classification of natural rock images using classifier combinations

Color is an essential feature that describes the image content and therefore colors occurring in the images should be effectively characterized in image classification. The selection of the number of the quantization levels is an important matter in the color description. On the other hand, when color representations using different quantization levels are combined, more accurate multilevel color description can be achieved. In this paper, we present a novel approach to multilevel color description of natural rock images. The description is obtained by combining separate base classifiers that use image histograms at different quantization levels as their inputs. The base classifiers are combined using classification probability vector (CPV) method that has proved to be an accurate way of combining classifiers in image classification.