Orientation Subbands Research Articles

Children are sensitive to mutual information in intermediate-complexity face and non-face features.

Understanding developmental changes in children's use of specific visual information for recognizing object categories is essential for understanding how experience shapes recognition. Research on the development of face recognition has focused on children's use of low-level information (e.g. orientation sub-bands), or high-level information. In face categorization tasks, adults also exhibit sensitivity to intermediate complexity features that are diagnostic of the presence of a face. Do children also use intermediate complexity features for categorizing faces and objects, and, if so, how does their sensitivity to such features change during childhood? Intermediate-complexity features bridge the gap between low- and high-level processing: they have computational benefits for object detection and segmentation, and are known to drive neural responses in the ventral visual system. Here, we have investigated the developmental trajectory of children's sensitivity to diagnostic category information in intermediate-complexity features. We presented children (5–10 years old) and adults with image fragments of faces (Experiment 1) and cars (Experiment 2) varying in their mutual information, which quantifies a fragment's diagnosticity of a specific category. Our goal was to determine whether children were sensitive to the amount of mutual information in these fragments, and if their information usage is different from adults. We found that despite better overall categorization performance in adults, all children were sensitive to fragment diagnosticity in both categories, suggesting that intermediate representations of appearance are established early in childhood. Moreover, children's usage of mutual information was not limited to face fragments, suggesting the extracting intermediate-complexity features is a process that is not specific only to faces. We discuss the implications of our findings for developmental theories of face and object recognition.

Low-level orientation information for social evaluation in face images.

Observers make a range of social evaluations based on facial appearance, including judgments of trustworthiness, warmth, competence, and other aspects of personality. What visual information do people use to make these judgments? While links have been made between perceived social characteristics and other high-level properties of facial appearance (e.g., attractiveness, masculinity), there has been comparatively little effort to link social evaluations to low-level visual features, like spatial frequency and orientation sub-bands, known to be critically important for face processing. We explored the extent to which different social evaluations depended critically on horizontal orientation energy vs. vertical orientation energy, as is the case for face identification and emotion recognition. We found that while trustworthiness judgments exhibited this bias for horizontal orientations, competence and dominance did not, suggesting that social evaluations may depend on a multi-channel representation of facial appearance at early stages of visual processing.

Body emotion recognition disproportionately depends on vertical orientations during childhood

Children’s ability to recognize emotional expressions from faces and bodies develops during childhood. However, the low-level features that support accurate body emotion recognition during development have not been well characterized. This is in marked contrast to facial emotion recognition, which is known to depend upon specific spatial frequency and orientation sub-bands during adulthood, biases that develop during childhood. Here, we examined whether children’s reliance on vertical vs. horizontal orientation energy for recognizing emotional expressions in static images of bodies changed during middle childhood (5 to 10 years old). We found that while children of all ages had an adult-like bias favoring vertical orientation energy, this effect was larger at younger ages. We conclude that in terms of information use, a key feature of the development of emotion recognition is improved performance with sub-optimal features for recognition – that is, learning to use less diagnostic features of the image is a slower process than learning to use more useful features.

Lithium-Ion Battery Capacity Estimation: A Method Based on Visual Cognition

This study introduces visual cognition into Lithium-ion battery capacity estimation. The proposed method consists of four steps. First, the acquired charging current or discharge voltage data in each cycle are arranged to form a two-dimensional image. Second, the generated image is decomposed into multiple spatial-frequency channels with a set of orientation subbands by using non-subsampled contourlet transform (NSCT). NSCT imitates the multichannel characteristic of the human visual system (HVS) that provides multiresolution, localization, directionality, and shift invariance. Third, several time-domain indicators of the NSCT coefficients are extracted to form an initial high-dimensional feature vector. Similarly, inspired by the HVS manifold sensing characteristic, the Laplacian eigenmap manifold learning method, which is considered to reveal the evolutionary law of battery performance degradation within a low-dimensional intrinsic manifold, is used to further obtain a low-dimensional feature vector. Finally, battery capacity degradation is estimated using the geodesic distance on the manifold between the initial and the most recent features. Verification experiments were conducted using data obtained under different operating and aging conditions. Results suggest that the proposed visual cognition approach provides a highly accurate means of estimating battery capacity and thus offers a promising method derived from the emerging field of cognitive computing.

Body emotion recognition depends on vertical orientation subbands during middle childhood

Body emotion recognition depends on vertical orientation subbands during middle childhood

Motion Estimation Based on Mutual Information and Adaptive Multi-Scale Thresholding.

This paper proposes a new method of calculating a matching metric for motion estimation. The proposed method splits the information in the source images into multiple scale and orientation subbands, reduces the subband values to a binary representation via an adaptive thresholding algorithm, and uses mutual information to model the similarity of corresponding square windows in each image. A moving window strategy is applied to recover a dense estimated motion field whose properties are explored. The proposed matching metric is a sum of mutual information scores across space, scale, and orientation. This facilitates the exploitation of information diversity in the source images. Experimental comparisons are performed amongst several related approaches, revealing that the proposed matching metric is better able to exploit information diversity, generating more accurate motion fields.

Orientation biases for facial emotion recognition in early childhood and adulthood

Face recognition in adults relies on information in specific spatial frequency and orientation subbands. Depending on the task, adults will perform better or worse based on what information is present in face images. An example of such an information bias is adults’ reliance on horizontal orientation energy for recognition; adults are generally able to recognize a face when horizontal information is retained and have substantial difficulty identifying a face when primarily vertical information is presented. While several recent studies have demonstrated various ways that horizontal orientation energy contributes to adult face processing, there have been as yet no studies describing how this bias emerges developmentally. Currently, we recruited participants between the ages of 5 and 6-years-old (N=16) to perform a simple emotion categorization task using face images with either primarily horizontal orientation energy, primarily vertical orientation energy, or both orientation subbands. A group of adults (N=18) were also run through the same task to determine the extent to which horizontal and vertical orientation energy contributed to recognition as a function of age. Specifically, we wished to determine if an adult-like bias for horizontal orientation energy is present early in childhood, or if this bias develops over the course of middle childhood. Our results demonstrated that both groups exhibited a horizontal orientation energy bias, though this bias differed as a function of age. Both groups were better able to classify facial emotion when horizontal features were retained than when only vertical features were available. However, children performed at chance when presented with vertically-filtered images, apparently unable to use sub-optimal features for emotion recognition. By comparison, adults’ performance was worse for vertically-filtered images, but still well above chance levels. We therefore propose that one feature of visual development may be the capability to use sub-optimal or weakly diagnostic information to support recognition. Meeting abstract presented at VSS 2015

Orientation biases for facial emotion recognition during childhood and adulthood

Facial emotion recognition develops slowly, with continuing changes in performance observable up to 10years of age and beyond. In the current study, we chose to examine how the use of specific low-level visual features for emotion recognition may change during childhood. Adults exhibit information biases for face recognition; specific spatial frequency and orientation sub-bands make a larger contribution to recognition than others. This means that depending on the specific task (e.g., identification, emotion recognition), participants will perform worse when some features are removed from the original image and better when those features are included. One example of such an information bias for face recognition is the differential contribution of horizontal orientation energy relative to vertical orientation energy; adult participants are better able to recognize faces and categorize their emotional expressions when horizontal information is included than when only vertical information is included. Although several recent studies have demonstrated various ways in which horizontal orientation energy (and so-called “bar-codes” for face appearance) contribute to adult face processing, there have been as yet no studies describing how such a bias emerges developmentally that may offer insight into the mechanisms underlying the slow development of facial emotion recognition. In the current study, we compared children’s (5- and 6-year-olds and 7- and 8-year-olds) and adults’ performance in a simple emotion categorization task using orientation-filtered faces to determine the extent to which horizontal and vertical orientation energy contributed to recognition as a function of age. We found that although all three participant groups exhibited a clear bias favoring the use of horizontal orientation energy, the nature of this bias differed as a function of age. Specifically, 5- and 6-year-olds exhibited a disproportionate performance cost when vertical orientation energy was all that was available relative to when stimuli were limited to horizontal orientation energy. One feature of the development of facial emotion recognition, thus, appears to be the capability to use suboptimal or weakly diagnostic information to support recognition.

Face and body emotion recognition depend on different orientation sub-bands

ABSTRACTFace and body perception are both disrupted by picture-plane inversion—180-degree rotation leads to poorer recognition and discrimination. For face stimuli, this inversion effect is carried by horizontally-oriented structures, which also carry more information for identity than vertical orientations. We examined whether face and body emotion recognition exhibited similar dependencies on low-level orientation energy (horizontal vs. vertical features) and whether the inversion effect for faces and bodies expressing different emotional expressions is carried by different orientation sub-bands. We asked observers to classify happy/sad faces and bodies that were filtered to include only horizontal orientation energy, only vertical orientation energy, or both orientations. We found that face and body emotion recognition rely on different low-level features—faces required horizontal features, while bodies required vertical features. In neither case was the inversion effect limited to the optimal band for emotion recognition. We conclude that face and body emotion recognition processes are tuned to different low-level features, but that in both cases the inversion effect (and any presumed “special” face or body processing associated with it) is applied fairly broadly to images in each category.

MULTIORIENTATION-BASED MULTISTRUCTURE MORPHOLOGICAL INPAINTING

A multiorientation-based multistructure morphological inpainting for the recovery of damaged digitized photographs is proposed. As the inpainting order plays a vital role for human visualization, the method is guided by the orientation of edges at the surrounding known regions of the missing (spoiled) domain. The damaged picture is decomposed into its constituent orientation subbands by steerable filters. The subband information is used for reconstructing the regions within the missing part at a particular orientation, as well as for guiding the integration of the reconstructed regions. Subbands having response at the boundary of the missing domain are named as constructive subbands. The damaged regions are morphologically eroded using the structuring elements of corresponding orientations that of the constructive subbands. The resultant image is obtained using winner-takes-all integration. The novelty of our approach is to explicitly specify the direction of filling thereby ensuring ease in convergence in different orientations and then streamlining the integration process to guarantee complete and natural look. Implementation of region-filling through morphological erosion, a noniterative and nonsampling process, makes the method faster than many other traditional texture synthesis inpainting algorithms and it successfully recovers images with better peak signal to noise ratios and structural similarity index even for massive damages.

Color contrast enhancement method using steerable pyramid transform

A new method of contrast enhancement based on steerable pyramid transform is presented in this work. The use of steerable filters is motivated by the fact that the images are to be observed by human and therefore it would be better to incorporate some knowledge on the Human Visual System in the design of the image processing tool. Here, the frequency and directional selectivity of the HVS is modeled by the steerable filters. The contrast is amplified using a selective nonlinear function which simulates the nonlinearity response of the HVS to the luminance stimuli. So the basic idea is to enhance the luminance signal irrespective of the two chrominance components using a multidirectional and multiscale decorrelation color transform. Initially the rgb (red, green and blue) color image is converted to lab (luminance and chrominance) color image. Only the luminance component is transformed by the steerable pyramid transform, so that the luminance component is independently decomposed into different scale and orientation sub-bands. The contrast in each sub-band is enhanced using a nonlinear mapping function. Finally the rgb color image is obtained from the enhanced luminance component along with the original chrominance components. The performance of the proposed method is objectively evaluated using spectrum energy analysis and a visibility map based on a perceptual filtering model. The obtained results confirm the efficiency of the method in enhancing subtle details without affecting color balance and without the usual noise amplification and edge ringing effect.

Multiresolution Denoising for Optical Coherence Tomography: A Review and Evaluation

Recently emerging non-invasive imaging modality - optical coherence tomography (OCT) – is becoming an increasingly important diagnostic tool in various medical applications. One of its main limitations is the presence of speckle noise which obscures small and low-intensity features. The use of multiresolution techniques has been recently reported by several authors with promising results. These approaches take into account the signal and noise properties in different ways. Approaches that take into account the global orientation properties of OCT images apply accordingly different level of smoothing in different orientation subbands. Other approaches take into account local signal and noise covariances. So far it was unclear how these different approaches compare to each other and to the best available single-resolution despeckling techniques. The clinical relevance of the denoising results also remains to be determined. In this paper we review systematically recent multiresolution OCT speckle filters and we report the results of a comparative experimental study. We use 15 different OCT images extracted from five different three-dimensional volumes, and we also generate a software phantom with real OCT noise. These test images are processed with different filters and the results are evaluated both visually and in terms of different performance measures. The results indicate significant differences in the performance of the analyzed methods. Wavelet techniques perform much better than the single resolution ones and some of the wavelet methods improve remarkably the quality of OCT images.

Rotation-Invariant Texture Retrieval via Signature Alignment Based on Steerable Sub-Gaussian Modeling

This paper addresses the construction of a novel efficient rotation-invariant texture retrieval method that is based on the alignment in angle of signatures obtained via a steerable sub-Gaussian model. In our proposed scheme, we first construct a steerable multivariate sub-Gaussian model, where the fractional lower-order moments of a given image are associated with those of its rotated versions. The feature extraction step consists of estimating the so-called covariations between the orientation subbands of the corresponding steerable pyramid at the same or at adjacent decomposition levels and building an appropriate signature that can be rotated directly without the need of rotating the image and recalculating the signature. The similarity measurement between two images is performed using a matrix-based norm that includes a signature alignment in angle between the images being compared, achieving in this way the desired rotation-invariance property. Our experimental results show how this retrieval scheme achieves a lower average retrieval error, as compared to previously proposed methods having a similar computational complexity, while at the same time being competitive with the best currently known state-of-the-art retrieval system. In conclusion, our retrieval method provides the best compromise between complexity and average retrieval performance.