Houssem Chatbri

Using Scale Space Filtering to Make Thinning Algorithms Robust Against Noise in Sketch Images

We apply scale space filtering to thinning of binary sketch images by introducing a framework for making thinning algorithms robust against noise. Our framework derives multiple representations of an input image within multiple scales of filtering. Then, the filtering scale that gives the best trade-off between noise removal and shape distortion is selected. The scale selection is done using a performance measure that detects extra artifacts (redundant branches and lines) caused by noise and shape distortions introduced by high amount of filtering. In other words, our contribution is an adaptive preprocessing, in which various thinning algorithms can be used, and which task is to estimate automatically the optimal amount of filtering to deliver a neat thinning result. Experiments using five state-of-the-art thinning algorithms, as the framework’s thinning stage, show that robustness against various types of noise was achieved. They are mainly contour noise, scratch, and dithers. In addition, application of the framework in sketch matching shows its usefulness as a preprocessing and normalization step that improves matching performances.

Sketch-Based Image Retrieval by Size-Adaptive and Noise-Robust Feature Description

We review available methods for Sketch-Based Image Retrieval (SBIR) and we discuss their limitations. Then, we present two SBIR algorithms: The first algorithm extracts shape features by using support regions calculated for each sketch point, and the second algorithm adapts the Shape Context descriptor to make it scale invariant and enhances its performance in presence of noise. Both algorithms share the property of calculating the feature extraction window according to the sketch size. Experiments and comparative evaluation with state-of-the-art methods show that the proposed algorithms are competitive in distinctiveness capability and robust against noise.

A comparative study using contours and skeletons as shape representations for binary image matching

An empirical study to compare contours and skeletons is conducted.Different image datasets are prepared and different image variations are generated.Results show the superiority of contours over skeletons.A noteworthy finding is the improvements of skeletons in the presence of noise. Contours and skeletons are well-known shape representations that embody visual information by using a limited set of object points. Both representations have been applied in various pattern recognition applications, while studies in cognitive science have investigated their roles in human perception. Despite their importance has been shown in the above-mentioned fields, to our knowledge no existing studies have been conducted to compare their performances. Filling this gap, this paper is an empirical study of these two shape representations by comparing their performances over different binary image categories and variations. The image categories include thick, elongated, and nearly thin images. Image variations include addition of noise to the contours, blurring, and size reduction. The comparative evaluation is achieved by resorting to object classification (OC) and content-based image retrieval (CBIR) algorithms and evaluation metrics. The main findings highlight the superiority of contours but the improvements observed when skeletons are used for images with noisy contours.

Sketch-based image retrieval by shape points description in support regions

We introduce a statistical shape descriptor for Sketch-Based Image Retrieval. The proposed descriptor combines feature information in near and far support regions defined for each sketch point. Two feature values are extracted from each point, corresponding to near and far support regions from the points perspective, and used to populate a 2-D histogram representing the shape features of the sketch image. The boundary between the support regions is calculated accordingly to each sketch point, which makes the approach scale invariant. We report results of objective evaluation of the proposed approach regarding robustness against noise, and comparative evaluation with three state-of-the-art methods, using an image database of scanned handwritten alphabets, digits, mathematical symbols and expressions. Experimental results show that the proposed method has competitive distinctiveness and robustness against noise.

Shape matching using keypoints extracted from both the foreground and the background of binary images

We introduce a descriptor for shape feature extraction and matching using keypoints that are extracted from both the foreground and the background of binary images. First, distance transform (DT) is applied on the image after contour detection. Then, connected components (CCs) of pixels having the same intensity are extracted. Keypoints correspond to centers of mass of CCs. A keypoint filtering mechanism is applied by estimating the spatial stability of keypoints when successive iterations of image blurring and binarization are applied. Finally, features are extracted for each keypoint using a round layout which radius is set depending on the keypoints location. We evaluate our descriptor using datasets of silhouette images, handwritten math expressions, and logos. Experimental results show that our descriptor is competitive compared with state-of-the-art methods, and that keypoint filtering is effective in reducing the number of keypoints without compromising matching performances.

A modular approach for query spotting in document images and its optimization using genetic algorithms

Query spotting in document images is a subclass of Content-Based Image Retrieval (CBIR) algorithms concerned with detecting occurrences of a query in a document image. Due to noise and complexity of document images, spotting can be a challenging task and easily prone to false positives and partially incorrect matches, thereby reducing the overall precision of the algorithm. A robust and accurate spotting algorithm is essential to our current research on sketch-based retrieval of digitized lecture materials. We have recently proposed a modular spotting algorithm in [1]. Compared to existing methods, our algorithm is both application-independent and segmentation-free. However, it faces the same challenges of noise and complexity of images. In this paper, inspired by our earlier research on optimizing parameter settings for CBIR using an evolutionary algorithm [2][3], we introduce a Genetic Algorithm-based optimization step in our spotting algorithm to improve each spotting result. Experiments using an image dataset of journal pages reveal promising performance, in that the precision is significantly improved but without compromising the recall of the overall spotting result.