Search

In this letter, we propose soft compression, an lossless compression approach to shape coding for images using location index and codebook of designed shapes with various sizes. This method is different from traditional image compression methods, as it aims at finding the basic shape blocks of pictures to improve the compression ratio from the perspective of information theory and frequency. In particular, the triplets including location and codeword are stored when encoding an image instead of the entire array. It is shown that soft compression can greatly reduce the bandwidth and storage space needed in the process of transmitting and storing the same kind of images.

Object recognition relies on different transformations of the retinal input, carried out by the visual system, that range from local contrast to object shape and category. While some of those transformations are thought to occur at specific stages of the visual hierarchy, the features they represent are correlated (e.g., object shape and identity) and selectivity for the same feature overlaps in many brain regions. This may be explained either by collinearity across representations or may instead reflect the coding of multiple dimensions by the same cortical population. Moreover, orthogonal and shared components may differently impact distinctive stages of the visual hierarchy. We recorded functional MRI activity while participants passively attended to object images and employed a statistical approach that partitioned orthogonal and shared object representations to reveal their relative impact on brain processing. Orthogonal shape representations (silhouette, curvature, and medial axis) independently explained distinct and overlapping clusters of selectivity in the occitotemporal and parietal cortex. Moreover, we show that the relevance of shared representations linearly increases moving from posterior to anterior regions. These results indicate that the visual cortex encodes shared relations between different features in a topographic fashion and that object shape is encoded along different dimensions, each representing orthogonal features.NEW & NOTEWORTHY There are several possible ways of characterizing the shape of an object. Which shape description better describes our brain responses while we passively perceive objects? Here, we employed three competing shape models to explain brain representations when viewing real objects. We found that object shape is encoded in a multidimensional fashion and thus defined by the interaction of multiple features.

The shape of a geospatial object is an important characteristic and a significant factor in spatial cognition. Existing shape representation methods for vector-structured objects in the map space a...

Fiber-reinforced composites have become the material of choice for aerospace structures because of their favorable strength-to-weight ratio. Given the increasing amounts of counterfeit composite parts showing up in the complex aerospace supply chain, it is absolutely vital to track a composite part throughout its lifecycle-from production to usage and to disposal. Existing barcoding methods are invasive, affect the structural properties of composites, and/or are vulnerable to tampering. We describe a universal method to store information in fiber-reinforced composites based on solid-state in situ reduction leading to embedded nanoparticles with controlled morphologies. This simple, cost-effective, mild, surfactant-free, and one-step protocol for the fabrication of embedded platinum nanostructures leads to morphology-based barcodes for polymeric composites. We also describe a coding methodology wherein a 1 × 1 cm code can represent 3.4 billion parts to 95 trillion parts, depending on the resolution required along with access to morphology-based chemical encryption systems.

American sign language recognition is still a research focus in computer vision community. Recently, most researches mainly extract low-level features for hand gesture recognition. These approaches perform poorly on recognizing gestures posed like a fist. In this paper, we propose a novel multimodal framework for sign language recognition system which exploits the Leap Motion Controller (LMC) and a webcam. We compute two sets of features. The first set is the angles at hand joints acquired by the LMC sensor. When, hand poses like a fist, the positions of the thumb joints captured by the LMC are not very precise. So, we should incorporate the second set of features extracted from the hand shape contour provided by a webcam. In this paper, we introduce a new mid-level feature, called Contour Segment Code (CSC), to represent hand shape contour. The proposed shape representation, first, extracts meaningful landmarks from the hand shape contour. CSC then encodes different segments of the hand contour into a code based on the shape landmarks. The extracted landmarks precisely determine the hand direction. The proposed method is tested by creating a very challenging dataset composed of 64,000 samples. Our experiments study the performance of the LMC and characteristics of CSC in different scenarios. The experimental results demonstrate the privileged performance of the proposed method against the systems which use depth images.

Shape is a popular way to define objects and shape coding is a key technique for object-based 3D video applications. In this paper, the issue of efficient shape coding for object-based 3D video applications is addressed, and a novel contour-based and chain-represented scheme is proposed. For a given 3D shape video, contour extraction and preprocessing are first implemented followed by chain-based representation. Then, to achieve high coding efficiency, a chain-based prediction and compensation technique is developed based on joint motion-compensated prediction and disparity-compensated prediction to effectively exploit the intra-view temporal correlation and the inter-view spatial correlation. Experiments are conducted, and the results demonstrate that the proposed scheme is more efficient than the existing methods, including state-of-the-art methods.

How to create a portable and quick way to detect multiple coexisting toxins is closely related to everyone's health. In this paper, we have established a real-time mycotoxin detection system that combined shape-encoded hydrogel particle preparation technology and image processing technology with smartphone portable devices. First, hydrogel microparticles containing a specific recognition toxin aptamer were programmable synthesized by stop-flow lithography. The hydrogel particles prepared by us had clear, variable signals and high coding capacity. Then, the indirect competitive detection based on aptamers was simple and rapid; the total reaction time was no more than 1 h 45 min and the image processing process was no more than 10 s. Finally, images could be captured by cameras on portable devices and smartphones. The self-built Android app that used the image recognition program installed on the smartphone would analyze the image and return the results in real time. The results showed that the detection limit reached 0.1 ng/mL, which was lower than the standard. In summary, this platform provides a fast, portable, high-throughput detection solution for real-time detection of mycotoxins, with excellent application prospects.

In the paper we study the recall efficiency in operators working with visual information: ability to remember geometric shapes and colors. We conducted an experiment in which an operator had to process the visually presented information in short-term memory in order to reproduce it in the future. The results of the experiment showed equal probability of errors when recalling colors and shapes. We also managed to distinguish two groups of geometric shapes in which there is high probability of interference. The first group consists: triangle, four-pointed star, and five-pointed star; the second: circle, rectangle, oval, and square.

Compression of point clouds has so far been confined to coding the positions of a discrete set of points in space and the attributes of those discrete points. We introduce an alternative approach based on volumetric functions, which are functions defined not just on a finite set of points, but throughout space. As in regression analysis, volumetric functions are continuous functions that are able to interpolate values on a finite set of points as linear combinations of continuous basis functions. Using a B-spline wavelet basis, we are able to code volumetric functions representing both geometry and attributes. Attribute compression is addressed in Part I of this paper, while geometry compression is addressed in Part II. Geometry is represented implicitly as the level set of a volumetric function (the signed distance function or similar). Experimental results show that geometry compression using volumetric functions improves over the methods used in the emerging MPEG Point Cloud Compression (G-PCC) standard.

We easily recognize objects across changes in viewpoint, but the underlying features are unknown. Here, we show that view invariance in monkey inferotemporal cortex is driven mainly by external object contours and is not specialized for object orientation. We also find that the responses to natural objects match with that of their silhouettes early in the response, and with inverted versions later in the response—indicative of a coarse-to-fine processing sequence in the brain., We effortlessly recognize objects across changes in viewpoint, but we know relatively little about the features that underlie viewpoint invariance in the brain. Here, we set out to characterize how viewpoint invariance in monkey inferior temporal (IT) neurons is influenced by two image manipulations—silhouetting and inversion. Reducing an object into its silhouette removes internal detail, so this would reveal how much viewpoint invariance depends on the external contours. Inverting an object retains but rearranges features, so this would reveal how much viewpoint invariance depends on the arrangement and orientation of features. Our main findings are 1) view invariance is weakened by silhouetting but not by inversion; 2) view invariance was stronger in neurons that generalized across silhouetting and inversion; 3) neuronal responses to natural objects matched early with that of silhouettes and only later to that of inverted objects, indicative of coarse-to-fine processing; and 4) the impact of silhouetting and inversion depended on object structure. Taken together, our results elucidate the underlying features and dynamics of view-invariant object representations in the brain. NEW & NOTEWORTHY We easily recognize objects across changes in viewpoint, but the underlying features are unknown. Here, we show that view invariance in the monkey inferotemporal cortex is driven mainly by external object contours and is not specialized for object orientation. We also find that the responses to natural objects match with that of their silhouettes early in the response, and with inverted versions later in the response—indicative of a coarse-to-fine processing sequence in the brain.

Background: Shape Coding is a visual coding system that has been used to teach English syntax and morphology to school-aged children with language impairment but has the potential to support the language output of people with aphasia. While visual coding has been used effectively in a number of studies targeting basic sentence structure, these approaches are difficult to expand to include more than a limited number of arguments or to encourage individuals to produce more complex sentences. Shape Coding allows the user to work with more complex structures and verb morphology and may be valuable in improving awareness of sentence structure in adults with acquired agrammatism.Aims: The aim of the current study is to investigate whether Shape Coding could improve the verbal output of two adult chronically agrammatic speakers.Methods & Procedures: The study involves two men with chronic non-fluent aphasia, one of whom had previously worked with Shape Coding. Repeated baseline measures were collected th...

Object recognition relies on different transformations of the retinal input, carried out by the visual system, that range from local contrast to object shape and category. While some of those transformations are thought to occur at specific stages of the visual hierarchy, the features they represent are correlated (e.g., object shape and identity) and selectivity for the same feature overlaps in many brain regions. This may be explained either by collinearity across representations, or may instead reflect the coding of multiple dimensions by the same cortical population. Moreover, orthogonal and shared components may differently impact on distinctive stages of the visual hierarchy. We recorded functional MRI (fMRI) activity while participants passively attended to object images and employed a statistical approach that partitioned orthogonal and shared object representations to reveal their relative impact on brain processing. Orthogonal shape representations (silhouette, curvature and medial-axis) independently explained distinct and overlapping clusters of selectivity in occitotemporal (OTC) and parietal cortex. Moreover, we show that the relevance of shared representations linearly increases moving from posterior to anterior regions. These results indicate that the visual cortex encodes shared relations between different features in a topographic fashion and that object shape is encoded along different dimensions, each representing orthogonal features.New & NoteworthyThere are several possible ways of characterizing the shape of an object. Which shape description better describes our brain responses while we passively perceive objects? Here, we employed three competing shape models to explain brain representations when viewing real objects. We found that object shape is encoded in a multi-dimensional fashion and thus defined by the interaction of multiple features.

Our visual abilities are unsurpassed because of a sophisticated code for objects located in the inferior temporal (IT) cortex. This code has remained a mystery because IT neurons show extremely diverse shape selectivity with no apparent organizing principle. Here, we show that there is an intrinsic component to selectivity in IT neurons. We tested IT neurons on distinct shapes and their parametric variations and asked whether neurons selective along one dimension were also selective along others. Selective neurons responded to fewer shapes and were narrowly tuned to local variations of these shapes, both along arbitrary morph lines and along variations in size, position, or orientation. For a subset of neurons, selective neurons were selective for both shape and texture. Finally, selective neurons were also more invariant in that they preserved their shape preferences across changes in size, position, and orientation. These observations indicate that there is an intrinsic constraint on the sharpness of tuning for the features coded by each IT neuron, making it always sharply tuned or always broadly tuned along all dimensions. We speculate that this may be an organizing principle throughout visual cortex.

Biological vision relies on representations of the physical world at different levels of complexity. Relevant features span from simple low-level properties, as contrast and spatial frequencies, to object-based attributes, as shape and category. However, how these features are integrated into coherent percepts is still debated. Moreover, these dimensions often share common biases: for instance, stimuli from the same category (e.g., tools) may have similar shapes. Here, using magnetoencephalography, we revealed the temporal dynamics of feature processing in human subjects attending to pictures of items pertaining to different semantic categories. By employing Relative Weights Analysis, we mitigated collinearity between model-based descriptions of stimuli and showed that low-level properties (contrast and spatial frequencies), shape (medial-axis) and category are represented within the same spatial locations early in time: 100-150ms after stimulus onset. This fast and overlapping processing may result from independent parallel computations, with categorical representation emerging later than the onset of low-level feature processing, yet before shape coding. Categorical information is represented both before and after shape also suggesting a role for this feature in the refinement of categorical matching.

Transform coding of arbitrary shape object is one of the core technologies of MPEG-4 video coding. In this chapter, we propose a shape adaptive Wavelet-Based Contourlet Transform framework for MPEG4 texture coding. In the shape adaptive texture coding framework, Wavelet-Based Contourlet Transform is divided into two parts: the Wavelet-Based Contourlet Transform of the shape mask, and the Wavelet-Based Contourlet Transform of the object texture. In the Wavelet-Based Contourlet Transform phase, the shape coding is carried out using lazy Wavelet-Based Contourlet Transform, and the texture coding is carried out by 9/7 wavelets in the form of lifting. In the zerotree quantization stage of wavelet coefficients, three quantization methods are adopted according to different needs. The experimental results show that the Wavelet-Based Contourlet Transform texture coding based on object has similar low complexity with the texture coding, such as SADCT, and the performance is better. Moreover, the coding effect is better than that of SADCT at low bit rate.

Submitted by Aline Batista (alinehb.ufpel@gmail.com) on 2019-03-18T18:20:24Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Dissertacao_Cassia_Cilene_Sampaio.pdf: 2768285 bytes, checksum: 446d98c0fbcd6078f8409f55741300c9 (MD5) Approved for entry into archive by Aline Batista (alinehb.ufpel@gmail.com) on 2019-03-19T12:54:04Z (GMT) No. of bitstreams: 2 Dissertacao_Cassia_Cilene_Sampaio.pdf: 2768285 bytes, checksum: 446d98c0fbcd6078f8409f55741300c9 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Made available in DSpace on 2019-03-19T12:54:12Z (GMT). No. of bitstreams: 2 Dissertacao_Cassia_Cilene_Sampaio.pdf: 2768285 bytes, checksum: 446d98c0fbcd6078f8409f55741300c9 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2018-08-31 Sem bolsa Nesta pesquisa buscou-se investigar o processo de ensino e aprendizagem da variação verbal por surdos em um ambiente de ensino de Português como segunda língua. No início do estudo os sujeitos da pesquisa demonstravam dificuldade de compreensão de como se organiza a escrita dos tempos verbais e, apresentavam reincidência do uso de verbos no infinitivo. Foi realizada uma pesquisa-ação, com aplicação de sequência didática que abordou o gênero narrativo. No processo de leitura e de realização dos exercícios foi utilizado o Sistema Shape Coding para explicar as diferenças entre os tempos verbais. O trabalho buscou privilegiar a experiência visual do surdo no processo de ensino e aprendizagem. A aplicação das sequencias didáticas ocorreu em sala de aula de escola bilíngue na qual a autora é docente. Os participantes são usuários de Libras como L1, todos com surdez profunda congênita e estudantes do ensino fundamental. O uso do sistema Shape Coding, por ser um recurso altamente visual, auxiliou e acelerou o processo de aprendizagem dos alunos participantes. Além disso, os princípios da abordagem comunicativa associados ao uso do Shape Coding demonstraram ser complementares dentro de um contexto pedagógico no qual a Libras é a língua de instrução. Com base nas análises dos exercícios e das produções escritas dos alunos participantes, sugere-se que o Shape Coding pode ser utilizado no ensino de língua escrita para surdos. This research aims to investigate the process of teaching and learning verbal variation by deaf people in a teaching Portuguese as a second language setting. At the beginning of the study, the individuals of the research demonstrated difficulty in understanding how the writing of verbal tenses is organized and showed a recurrent use of verbs in the infinitive. Action research was carried out, with the application of a didactic sequence that dealt with the narrative genre. In the process of reading and performing the activities, the Shape Coding System was used to explain the differences between verbal tenses. The work sought to privilege the visual experience of the deaf in the process of teaching and learning. The application of didactic sequences occurred in a bilingual school classroom in which the author is a teacher. Participants are users of Libras as L1, all with profound congenital deafness and elementary school students. The use of the Shape Coding system, as a highly visual resource, aided and accelerated the learning process of the participating students. In addition, communicative approach principles combined with Shape Coding have been shown to be complementary within a pedagogical context in which Libras is the language of instruction. Based on the analyzes of the exercises and the written productions of the participating students, it suggests that Shape Coding is useable in the teaching of written language for the deaf.

The technique of contour-based predictive shape coding is addressed.A highly efficient scheme for contour-based predictive shape coding is proposed.A new chain-based method of motion estimation and motion compensation is given.A new technique is developed for encoding the residuals. A new scheme for contour-based predictive shape coding is proposed aiming to acquire high coding efficiency, where the temporal correlations among object contours are effectively exploited. For a given binary shape image, the object contours are firstly extracted and thinned to be perfect single-pixel width followed by chain-based representation. Then a chain-based motion estimation and compensation technique is developed to remove temporal correlations among object contours to reduce the data to be encoded. Finally, by further exploiting the spatial correlations within chain links, a novel method is introduced to efficiently encode the residuals together with the motion displacements. Experiments are conducted and the results show that the proposed scheme is considerably more efficient than the existing techniques.

In this paper, we present a new model to measure the contour point distortion for the vertex-based shape coding with B-splines, called accurate distortion measurement using analytical model (ADMAM). Different from existing distortion measurements containing approximation, quantization or parameterization process, our distortion is defined on the original B-spline. It is modeled as the shortest distance of associated contour point from the original B-spline, which is in line with the subjective-based objective quality metric. The geometric relationships are introduced to simplify the model computation, followed by a hybrid admissible distortion checking algorithm to reduce the execution time. Our theoretical analysis and experimental results demonstrate that the ADMAM can lead to the smallest bit-rate among all the distortion measurements that guarantee the admissible distortion, when the operational rate-distortion optimal shape coding framework is applied. Moreover, if the original contour has NC points, it takes only O ( N C ) time for both peak and mean-squared segment distortion measuring paradigms, which is the lowest computational complexity among all the existing distortion measurements.

Text recognition in scene image and video frames is difficult because of low resolution, blur, background noise, etc. Since traditional OCRs do not perform well in such images, information retrieval using keywords could be an alternative way to index/retrieve such text information. Date is a useful piece of information which has various applications including date-wise videos/scene searching, indexing or retrieval. This paper presents a date spotting based information retrieval system for natural scene image and video frames where text appears with complex backgrounds. We propose a line based date spotting approach using hidden Markov model (HMM) which is used to detect the date information in a given text. Different date models are searched from a line without segmenting characters or words. Given a text line image in RGB, we apply an efficient gray image conversion to enhance the text information. Wavelet decomposition and gradient sub-bands are used to enhance text information in gray scale. Next, pyramid histogram of oriented gradient (PHOG) feature has been extracted from gray image and binary images for date-spotting framework. Binary and gray image features are combined by MLP based tandem approach. Finally, to boost the performance further, a shape coding based scheme is used to combine the similar shape characters in same class during word spotting. For our experiment, three different date models have been constructed to search similar date information having numeric dates that contains numeral values and punctuations and semi-numeric that contains dates with numerals along with months in scene/video text. We have tested our system on 1648 text lines and the results show the effectiveness of our proposed date spotting approach.

The majority of text summarization techniques in literature depend, in one way or another, on language dependent pre-structured lexicons, databases, taggers and/or parsers. Such techniques require a prior knowledge of the language of the text being summarized. In this paper we propose an extractive text summarization tool, UnB Language Independent Text Summarizer (UnB-LITS), which is capable of performing text summarization in a language independent manner. The new model depends on intrinsic characteristics of the text being summarized rather than its language and thus eliminates the need for language dependent lexicons, databases, taggers or parsers. Within this tool, we develop an innovative way of coding the shapes of text elements (words, n-grams, sentences and paragraphs), in addition to proposing language independent algorithms that is capable of normalizing words and performing relative stemming or lemmatization. The proposed algorithms and Shape-Coding routine enable the UnB-LITS tool to extract intrinsic features of document elements and score them statistically to extract a representative extractive summary independent of the document language. In this paper we focused on single document summarization of western European languages. The tool was tested on hundreds of documents written in English, Portuguese, French and Spanish and showed better performance as compared with the results obtained in literature as well as from commercial summarizers.

Retrieval of text information from natural scene images and video frames is a challenging task due to its inherent problems like complex character shapes, low resolution, background noise, etc. Available OCR systems often fail to retrieve such information in scene/video frames. Keyword spotting, an alternative way to retrieve information, performs efficient text searching in such scenarios. However, current word spotting techniques in scene/video images are script-specific and they are mainly developed for Latin script. This paper presents a novel word spotting framework using dynamic shape coding for text retrieval in natural scene image and video frames. The framework is designed to search query keyword from multiple scripts with the help of on-the-fly script-wise keyword generation for the corresponding script. We have used a two-stage word spotting approach using Hidden Markov Model (HMM) to detect the translated keyword in a given text line by identifying the script of the line. A novel unsupervised dynamic shape coding based scheme has been used to group similar shape characters to avoid confusion and to improve text alignment. Next, the hypotheses locations are verified to improve retrieval performance. To evaluate the proposed system for searching keyword from natural scene image and video frames, we have considered two popular Indic scripts such as Bangla (Bengali) and Devanagari along with English. Inspired by the zone-wise recognition approach in Indic scripts[1], zone-wise text information has been used to improve the traditional word spotting performance in Indic scripts. For our experiment, a dataset consisting of images of different scenes and video frames of English, Bangla and Devanagari scripts were considered. The results obtained showed the effectiveness of our proposed word spotting approach., Comment: Multimedia Tools and Applications, Springer

The novelty of this study was to throw light on physical properties or intrinsic features which influence the communication of a product character and the teaching the skill of manipulating such properties to product design students. The paper benchmarks successful strategies for shape coding: playful, simple, and modern characters. These characters were selected because existing literature has claimed them as appearance attributes (product characters) universally recognized by ordinary users (non-professional designers) and expert users alike, resulting in their higher success rate for meaning communication. Although existing literature has explored relationships between product characters and physical properties using statistical and mathematical tools, this research attempts to establish that relationship from a more practical and hands-on approach by having Designers in training (students) develop design strategies through product forms. Professional designers and theoretical academicians regard product semantics in fundamentally different ways, where, designers tend to depend more on intuition and experience seldom using qualitative tools when dealing with meaning communication through form giving. The teaching discussed here contributes to bridge this gap by introducing a design process to students (designers in training) based on language structures while shaping intrinsic features of a product form.

The paper mainly argues two questions, one is how to extract the feature of the ceramic productions based on the pairwise geometric histogram, and another is how to index and retrieval the ceramic productions. They are researched including the PGH-based shape coding, the prediction of the match cost, and the PGH-based retrieval. The PGH-based retrieval model of the ceramic productions is analyzed, proposed and applied to the ceramic retrieval experiment at last. The experimental results show that the PGH-based retrieval has the high retrieval accuracy to the same kind of target with different size or direction. It has a universal applicability.

Number is a complex category, as with the word "number" we may refer to different entities. First, it is a perceptual property that characterizes any set of individual items, namely its cardinality. The ability to extract the (approximate) cardinality of sets is almost universal in the animal domain and present in humans since birth. In primates, posterior parietal cortex seems to be a crucial site for this ability, even if the degree of selectivity of numerical representations in parietal cortex reported to date appears much lower compared to that of other semantic categories in the ventral stream. Number can also be intended as a mathematical object, which we humans use to count, measure, and order: a (verbal or visual) symbol that stands for the cardinality of a set, the intensity of a continuous quantity or the position of an item on a list. Evidence points to a convergence towards parietal cortex for the semantic coding of numerical symbols and to the bilateral occipitotemporal cortex for the shape coding of Arabic digits and other number symbols.

The sensitivity to illumination variations, pose, gender, age, clothing and any another source of changes, can be one of the most important challenges, in gait recognition system. In this paper, we adopt many approaches to extract signatures of human body (static model) using a model-based method, such as static body parameters, ellipse-fitting and robust shape coding. To reduce the dimension of this features set, a principal component analysis (PCA) technique is employed. Then, a deep belief networks classifier is used to classify the gait signatures. The performance of the deep belief network (DBN) is superior to other classifiers such as k-nearest neighbour (KNN) and dynamic times warping (DTW). The comparison is performed for viewpoint changes, clothing and carrying conditions. The proposed approach has been validated on the gait database B.

In its search for neural codes, the field of visual neuroscience has uncovered neural representations that reflect the structure of stimuli of variable complexity from simple features to object categories. However, accumulating evidence suggests an adaptive neural code that is dynamically shaped by experience to support flexible and efficient perceptual decisions. Here, we review work showing that experience plays a critical role in molding midlevel visual representations for perceptual decisions. Combining behavioral and brain imaging measurements, we demonstrate that learning optimizes feature binding for object recognition in cluttered scenes, and tunes the neural representations of informative image parts to support efficient categorical judgements. Our findings indicate that similar learning mechanisms may mediate long-term optimization through development, tune the visual system to fundamental principles of feature binding, and optimize feature templates for perceptual decisions.