A hybrid Retinex-based image enhancement algorithm is proposed to improve the quality of images captured by unmanned aerial vehicles (UAVs) in this paper. Hyperparameters of the employed multi-scale Retinex with chromaticity preservation (MSRCP) model are automatically tuned via a two-phase evolutionary computing algorithm. In the two-phase optimization algorithm, the Rao-2 algorithm is applied to performing the global search and a solution is obtained by maximizing the objective function. Next, the Nelder-Mead simplex method is used to improve the solution via local search. Real UAV-taken images of bad quality are collected to verify the performance of the proposed algorithm. Meanwhile, four famous image enhancement algorithms, Multi-Scale Retinex, Multi-Scale Retinex with Color Restoration, Automated Multi-Scale Retinex, and MSRCP are utilized as benchmarking methods. Meanwhile, two commonly used evolutionary computing algorithms, particle swarm optimization and flower pollination algorithm, are considered to verify the efficiency of the proposed method in tuning parameters of the MSRCP model. Experimental results demonstrate that the proposed method achieves the best performance compared with benchmarks and thus the proposed method is applicable for real UAV-based applications.

Faster R-CNN uses a region proposal network which consists of a single scale convolution filter and fully connected networks to localize detected regions. However, using a single scale filter is not enough to detect full regions of characters. In this letter, we propose a simple but effective way, i.e., utilizing variously sized convolution filters, to accurately detect Chinese characters of multiple scales in documents. We experimentally verified that our method improved IoU by 4% and detection rate by 3% than the previous single scale Faster R-CNN method.

In this letter, we propose a method for obtaining a clear and natural output image by tuning the illumination component in an input image. The proposed method is based on the retinex process and it is suitable for the image quality improvement of images of which illumination is insufficient.

We propose an edge-preserving multiscale image decomposition method using filters for non-equispaced signals. It is inspired by the domain transform, which is a high-speed edge-preserving smoothing method, and it can be used in many image processing applications. One of the disadvantages of the domain transform is sensitivity to noise. Even though the proposed method is based on non-equispaced filters similar to the domain transform, it is robust to noise since it employs a multiscale decomposition. It uses the Laplacian pyramid scheme to decompose an input signal into the piecewise-smooth components and detail components. We design the filters by using an optimization based on edge-preserving smoothing with a conversion of signal distances and filters taking into account the distances between signal intervals. In addition, we also propose construction methods of filters for non-equispaced signals by using arbitrary continuous filters or graph spectral filters in order that various filters can be accommodated by the proposed method. As expected, we find that, similar to state-of-the-art edge-preserving smoothing techniques, including the domain transform, our approach can be used in many applications. We evaluated its effectiveness in edge-preserving smoothing of noise-free and noisy images, detail enhancement, pencil drawing, and stylization.

Most unsupervised video segmentation algorithms are difficult to handle object extraction in dynamic real-world scenes with large displacements, as foreground hypothesis is often initialized with no explicit mutual constraint on top-down spatio-temporal coherency despite that it may be imposed to the segmentation objective. To handle such situations, we propose a multiscale saliency flow (MSF) model that jointly learns both foreground and background features of multiscale salient evidences, hence allowing temporally coherent top-down information in one frame to be propagated throughout the remaining frames. In particular, the top-down evidences are detected by combining saliency signature within a certain range of higher scales of approximation coefficients in wavelet domain. Saliency flow is then estimated by Gaussian kernel correlation of non-maximal suppressed multiscale evidences, which are characterized by HOG descriptors in a high-dimensional feature space. We build the proposed MSF model in accordance with the primary object hypothesis that jointly integrates temporal consistent constraints of saliency map estimated at multiple scales into the objective. We demonstrate the effectiveness of the proposed multiscale saliency flow for segmenting dynamic real-world scenes with large displacements caused by uniform sampling of video sequences.

This work presents an approximate global optimization method for image halftone by fusing multi-scale information of the tree model. We employ Gaussian mixture model and hidden Markov tree to characterized the intra-scale clustering and inter-scale persistence properties of the detailed coefficients, respectively. The model of multiscale perceived error metric and the theory of scale-related perceived error metric are used to fuse the statistical distribution of the error metric of the scale of clustering and cross-scale persistence. An Energy function is then generated. Through energy minimization via graph cuts, we gain the halftone image. In the related experiment, we demonstrate the superior performance of this new algorithm when compared with several algorithms and quantitative evaluation.

The current paper presents an effective deblocking algorithm for block-based coded images using singularity detection in a wavelet transform. Blocking artifacts appear periodically at block boundaries in block-based coded images. The local maxima of a wavelet transform modulus detect all singularities, including blocking artifacts, from multiscale edges. Accordingly, the current study discriminates between a blocking artifact and an edge by estimating the Lipschitz regularity of the local maxima and removing the wavelet transform modulus of a blocking artifact that has a negative Lipschitz regularity exponent. Experimental results showed that the performance of the proposed algorithm was objectively and subjectively superior.

In this paper, we propose an ECG waveform compression technique based on the matching pursuit. The matching pursuit is an iterative non-orthogonal signal expansion technique. A signal is decomposed to atoms in a function dictionary. The constraint to the dictionary is only the over-completeness to signals. The function dictionary can be defined to be best match to the structure of the ECG waveform. In this paper, we introduce the multiscale analysis to the implementation of inner product computations between signals and atoms in the matching pursuit iteration. The computational cost can be reduced by utilization of the filter bank of the multiscale analysis. We show the waveform approximation capability of the matching pursuit with multiscale analysis. We show that a simple 4-tap integer filter bank is enough to the approximation and compression of ECG waveforms. In ECG waveform compression, we apply the error feed-back procedure to the matching pursuit iteration to reduce the norm of the approximation error. Finally, actual ECG waveform compression by the proposed method are demonstrated. The proposed method achieve the compression by the factor 10 to 30. The compression ratio given by the proposed method is higher than the orthogonal wavelet transform coding in the range of the reconstruction precision lower than 9% in PRD.

In this paper, we have proposed a new method for diffuse liver disease classification with sonogram, including the normal liver, hepatitis and cirrhosis, from a new point of view "scale. " The new system utilizes a multiscale analysis tool, called wavelet transforms, to analyze the ultrasonic liver images. A new set of features consisting of second order statistics derived from the wavelet transformed images is employed. From these features, we have found that the third scale is the representative scale for the classification of the considered liver diseases, and the horizontal wavelet transform can improve the representation of the corresponding features. Experimental results show that our method can achieve about 88% correct classification rate which is superior to other measures such as the co-occurrence matrices, the Fourier power spectrum, and the texture spectrum. This implies that our feature set can access the granularity from sonogram more effectively. It should be pointed out that our features are powerful for discriminating the normal livers from the cirrhosis because there is no misclassification samples between the normal liver and the cirrhosis sets. In addition, the experimental results also verify the usefulness of "scale" because our multiscale feature set can gain eighteen percent advantage over the direct use of the statistical features. This means that the wavelet transform at proper scales can effectively increase the distances among the statistical feature clusters of different liver diseases.

Fuzzy rule-based edge detection using multiscale edge images is proposed. In this method, the edge image is obtained by fuzzy approximate reasoning from multiscale edge images which are obtained by derivative operators with various window sizes. The effect of utilizing multiscale edge images for edge detection is already known, but how to design the rules for deciding edges from multiscale edge images is not clarified yet. In this paper, the rules are represented in a fuzzy style, since edges are usually defined ambiguously, and the fuzzy rules are designed optimally by a training method. Here, the fuzzy approximate reasoning is expressed as a nonlinear function of the multiscale edge image data, and the nonlinear function is optimized so that the mean square error of the edge detection be the minimum. Computer simulations verify its high performance for actual images.

We introduce an image contour clustering method based on a multiscale image representation and its application to image compression. Multiscale gradient planes are obtained from the mean squared sum of 2D wavelet transform of an image. The decay on the multiscale gradient planes across scales depends on the Lipshitz exponent. Since the Lipshitz exponent indicates the spatial differentiability of an image, the multiscale gradient planes represent smoothness or sharpness around edges on image contours. We apply vector quatization to the multiscale gradient planes at contours, and cluster the contours in terms of represntative vectors in VQ. Since the multiscale gradient planes indicate the Lipshitz exponents, the image contours are clustered according to its gradients and Lipshitz exponents. Moreover, we present an image recovery algorithm to the multiscale gradient planes, and we achieve the skech-based image compression by the vector quantization on the multiscale gradient planes.

Antidiffusion is a process running the diffusion equation reversely in the time domain. Though extremely important for image restoration of the Gaussian blur, it is a horribly ill-posed problem, since minor noise leads to very erroneous results. To solve this ill-posed problem stably, in this paper we first apply a multiscale method to decompose images into various scale components using the Gaussian and Laplacian of Gaussian (LOG) filters. We then show that the restored images can be reconstructed from the components using shrunk Gaussian and LOG filters. Our algorithm has a closed form solution, and is robust to noise because it is performed by the integration computation (convolution), contrasting with the differential computation required by direct discretization of the antidiffusion equation. The antidiffusion algorithm is also computationally efficient since the convolution is row and column separable. Finally, a comparison between the algorithm and the well-known Wiener filter is conducted. Experiments show that our algorithm is really stable and images can be restored satisfactorily.

The pattern spectrum has been proposed to represent morphological size distribution of an image. However, the conventional pattern spectrum cannot extract approximate shape information from image objects spotted by noisy pixels since this is based only on opening. In this paper, a novel definition of the pattern spectrum, morphological multiresolution pattern spectrum (MPS), involving both opening and closing is proposed. MPS is capable of distinguishing details from approximate information of the image.