Toshiyuki DOBASHI Masahiro IWAHASHI Hitoshi KIYA
This letter considers a unified tone mapping operation (TMO) for HDR images. The unified TMO can perform tone mapping for various HDR image formats with a single common operation. The integer TMO which can realize unified tone mapping by converting an input HDR image into an intermediate format is proposed. This method can be executed efficiently with low memory and low performance processor. However, only floating-point HDR image formats have been considered in the method. In other words, a long-integer which is one of the HDR image formats has not been considered in the method. This letter applies the method to a long-integer format, and confirm its performance. The experimental results show the proposed method is effective for an integer format in terms of the resources such as the computational cost and the memory cost.
Fairoza Amira BINTI HAMZAH Taichi YOSHIDA Masahiro IWAHASHI Hitoshi KIYA
As three dimensional (3D) discrete wavelet transform (DWT) is widely used for high resolution volumetric data compression, and to further improve the performance of lossless coding, the adaptive directional lifting (ADL) structure based on non-separable 3D DWT with a (5,3) filter is proposed in this paper. The proposed 3D DWT has less lifting steps and better prediction performance compared to the existing separable 3D DWT with fixed filter coefficients. It also has compatibility with the conventional DWT defined by the JPEG2000 international standard. The proposed method shows comparable and better results with the non-separable 3D DWT and separable 3D DWT and it is effective for lossless coding of high resolution volumetric data.
Yuma KINOSHITA Sayaka SHIOTA Masahiro IWAHASHI Hitoshi KIYA
A number of successful tone mapping operators (TMOs) for contrast compression have been proposed due to the need to visualize high dynamic range (HDR) images on low dynamic range devices. This paper proposes a novel inverse tone mapping (TM) operation and a new remapping framework with the operation. Existing inverse TM operations require either the store of some parameters calculated in forward TM, or data-depended operations. The proposed inverse TM operation enables to estimate HDR images from LDR ones mapped by the Reinhard's global operator, not only without keeping any parameters but also without any data-depended calculation. The proposed remapping framework with the inverse operation consists of two TM operations. The first TM operation is carried out by the Reinhard's global operator, and then the generated LDR one is stored. When we want different quality LDR ones, the proposed inverse TM operation is applied to the stored LDR one to generate an HDR one, and the second TM operation is applied to the HDR one to generate an LDR one with desirable quality, by using an arbitrary TMO. This framework allows not only to visualize an HDR image on low dynamic range devices at low computing cost, but also to efficiently store an HDR one as an LDR one. In simulations, it is shown that the proposed inverse TM operation has low computational cost, compared to the conventional ones. Furthermore, it is confirmed that the proposed framework allows to remap the stored LDR one to another LDR one whose quality is the same as that of the LDR one remapped by the conventional inverse TMO with parameters.
Sakol UDOMSIRI Masahiro IWAHASHI Shogo MURAMATSU
This paper proposes a new type of layered video coding especially for the use of monitoring water level of a river. A sensor node of the system decomposes an input video signal into some kinds of component signals and produces a bit stream functionally separated into three layers. The first layer contains the minimum components effective for detecting the water level. It is transmitted at very low bit rate for regular monitoring. The second layer contains signals for thumb-nail video browsing. The third layer contains additional data for decoding the original video signal. These are transmitted in case of necessity. A video signal is decomposed into several bands with the three dimensional Haar transform. In this paper, optimum bands to be contained into the 1st layer are experimentally investigated considering both of water level detection and data size to be transmitted. As a result, bit rate for transmitting the first layer is reduced by 32.5% at the cost of negligible 3.7% decrease of recognition performance for one of video examples.
Isana FUNAHASHI Taichi YOSHIDA Xi ZHANG Masahiro IWAHASHI
In this paper, we propose an image adjustment method for multi-exposure images based on convolutional neural networks (CNNs). We call image regions without information due to saturation and object moving in multi-exposure images lacking areas in this paper. Lacking areas cause the ghosting artifact in fused images from sets of multi-exposure images by conventional fusion methods, which tackle the artifact. To avoid this problem, the proposed method estimates the information of lacking areas via adaptive inpainting. The proposed CNN consists of three networks, warp and refinement, detection, and inpainting networks. The second and third networks detect lacking areas and estimate their pixel values, respectively. In the experiments, it is observed that a simple fusion method with the proposed method outperforms state-of-the-art fusion methods in the peak signal-to-noise ratio. Moreover, the proposed method is applied for various fusion methods as pre-processing, and results show obviously reducing artifacts.
Pavol ZAVARSKY Takeshi MYOKEN Noriyoshi KAMBAYASHI Shinji FUKUMA Masahiro IWAHASHI
The paper shows some of benefits of multi-unitary decomposition in signal analysis applications. It is emphasized that decompositions of complex discrete-time signals onto a single basis provide an incomplete and in such way potentially misleading image of the signals in signal analysis applications. It is shown that the multi-unitary decimated filter banks which decompose the analyzed signal onto several bases of the given vector space can serve as a tool which provides a more complete information about the signal and at the same time the filter banks can enjoy efficient polyphase component implementation of maximally decimated, i. e. nonredundant, filter banks. An insight into the multi-unitary signal decomposition is provided. It is shown that the multiple-bases representation leads to an efficient computation of frequency domain representations of signals on a dense not necessarily uniform frequency grid. It is also shown that the multiple-bases representation can be useful in the detection of tones in digital implementations of multifrequency signaling, and in receivers of chirp systems. A proof is provided that there are possible benefits of the multiple-bases representations in de-noising applications.
Masahiro IWAHASHI Somchart CHOKCHAITAM Narong BUABTHONG Pavol ZAVARSKY Noriyoshi KAMBAYASHI
A new lossless scalable coding based on a lossless wavelet transform (LWT) and a lossless multi-channel prediction (LMP) is proposed. A rough image can be expanded from a part of the bit stream for the use of progressive transmission. The LMP, a non-separable 2D filter bank, is optimized for an arbitrary input image signal so that remaining correlation of the band signals of the LWT can be utilized. Filter coefficients are optimized for each of input images under the lossless coding gain.
Somchart CHOKCHAITAM Masahiro IWAHASHI Pavol ZAVARSKY Noriyoshi KAMBAYASHI
In this report, we propose an integrated lossy and lossless image coding, which is possible to be implemented on one architecture, based on combination of lossless wavelet transform (LWT) and lossy-lossless multi-channel prediction (LLMP). The LWT is applied to divide input signals into frequency subbands as octave-band decomposition, whereas the LLMP is designed as a non-separable two-dimensional filter bank including quantization step size and local decoding to enhance coding performance in both lossless coding and lossy coding. Its filter coefficients are determined to minimize total bit rate for lossless coding, and the optimum quantization step size is applied to maximize lossy coding gain. The local decoding is applied to avoid quantization error effect. The experimental results confirm effectiveness of our proposed method.
Yo UMEKI Taichi YOSHIDA Masahiro IWAHASHI
In this paper, we propose a method of salient object detection based on distributed seeds and a co-propagation of seed information. Salient object detection is a technique which estimates important objects for human by calculating saliency values of pixels. Previous salient object detection methods often produce incorrect saliency values near salient objects in the case of images which have some objects, called the leakage of saliencies. Therefore, a method based on a co-propagation, the scale invariant feature transform, the high dimensional color transform, and machine learning is proposed to reduce the leakage. Firstly, the proposed method estimates regions clearly located in salient objects and the background, which are called as seeds and resultant seeds, are distributed over images. Next, the saliency information of seeds is simultaneously propagated, which is then referred as a co-propagation. The proposed method can reduce the leakage caused because of the above methods when the co-propagation of each information collide with each other near the boundary. Experiments show that the proposed method significantly outperforms the state-of-the-art methods in mean absolute error and F-measure, which perceptually reduces the leakage.
Hiroaki IWAI Masahiro IWAHASHI Hitoshi KIYA
We propose two methods for avoiding the checkerboard distortion which is caused by finite word length error. The first method derives the bit length of filter coefficients required for avoiding the checkerboard distortion under a certain word length. In the second method, the checkerboard distortion can be avoided by using the cascade structure which consists of zero-hold kernel and a time-invariant filter factorized from a filter with structure for avoiding the checkerboard distortion under linear systems. It is demonstrated by simulations that we can avoid the checkerboard distortion by using these proposed methods.
Somchart CHOKCHAITAM Masahiro IWAHASHI Pavol ZAVARSKY Noriyoshi KAMBAYASHI
In this paper, we propose a bit-rate adaptive coding system based on lossless DCT (L-DCT). Our adaptive coding system consists of three different operation modes: lossless, near-lossless and lossy coding modes. Quantization is applied in transform domain (after the L-DCT) and spatial domain (before the L-DCT) in lossy mode and near-lossless mode, respectively. Our adaptive coding system can automatically select its operation mode at a given bit rate because it contains a function to calculate the turning point between near-lossless mode and lossy mode from characteristic of input signal. Existence of the turning point is mathematically proved in this paper. Simulation results confirm not only effectiveness of our adaptive coding system but also accuracy of our theoretical analysis.
Somchart CHOKCHAITAM Masahiro IWAHASHI
In this paper, we propose lossless/lossy coding criterion as a new objective criterion to theoretically evaluate coding performance of the lossless/lossy wavelet (LLW). The proposed lossless/lossy coding criterion consists of three parameters: "lossless coding criterion," "quantization-lossy coding gain" and "rounding errors. " The first parameter is a criterion to evaluate lossless coding performance of the LLW, whereas the second and the third parameters are criteria to evaluate lossy coding performance of the LLW at low bit rate and high bit rate, respectively. Relation among those three parameters is clearly illustrated in this paper. Performances of 15 kinds of the LLW are measured with two-dimensional (2D) octave-decomposition by applying some standard images and 2D AR(1) model as input signals.
Pavol ZAVARSKY Nobuo FUJII Masahiro IWAHASHI Noriyoshi KAMBAYASHI Shinji FUKUMA Takeshi MYOKEN
A simple but efficient method to improve readability of discrete pseudo time-frequency representations (TFRs) of nonstationary signals by the reassignment of the representations in discrete frequency dimension is presented. The method does not rely on the nonzero time derivative of the window function employed in the estimation of pseudo TFR. This property of the reassignment method is advantageous because the method can provide an improved readability in the situation when a known reassignment method is unefficient. The reassignment of the TFRs of corrupted signals is discussed. Numerical examples are included to illustrate the performance of the proposed method.
Masahiro IWAHASHI Hitoshi KIYA
This paper proposes a reversible two dimensional (2D) discrete wavelet transform (DWT) for lossless coding which is compatible with the irreversible 2D 9-7 DWT for lossy coding in the JPEG 2000. Since all the filters and scalings are factorized into a product of lifting steps, and signal values are rounded into integers, the proposed DWT is reversible and applicable to lossless coding of 2D signals. We replace a part of the separable 2D transfer function of the 2D DWT by a non separable 2D lifting structure, so that the number of rounding operations is decreased. We also investigate performance of the DWT under octave decomposition case and theoretically endorse it. As a result, reduction of the rounding errors due to the replacement was confirmed. It means that compatibility of the reversible DWT to the irreversible 2D 9-7 DWT is improved.
Somchart CHOKCHAITAM Masahiro IWAHASHI Somchai JITAPUNKUL
In this paper, we propose a new one-dimensional (1D) integer discrete cosine transform (Int-DCT) for unified lossless/lossy image compression. The proposed 1D Int-DCT is newly designed to reduce rounding effects by minimizing number of rounding operations. The proposed Int-DCT can be operated not only lossless coding for a high quality decoded image but also lossy coding for a compatibility with the conventional DCT-based coding system. Both theoretical analysis and simulation results confirm an effectiveness of the proposed Int-DCT.
Toshiyuki DOBASHI Tatsuya MUROFUSHI Masahiro IWAHASHI Hitoshi KIYA
A global tone mapping operation (TMO) for high dynamic range (HDR) images with fixed-point arithmetic is proposed and evaluated in this paper. A TMO generates a low dynamic range (LDR) image from an HDR image by compressing its dynamic range. Since an HDR image is generally expressed in a floating-point data format, a TMO also deals with floating-point data even though a resultant LDR image is integer data. The proposed method treats a floating-point number as two 8-bit integer numbers which correspond to an exponent part and a mantissa part, and applies tone mapping to these integer numbers separately. Moreover, the method conducts all calculations in the tone mapping with only fixed-point arithmetic. As a result, the method reduces a memory cost and a computational cost. The evaluation shows that the proposed method reduces 81.25% of memory usage. The experimental results show that the processing speed of the proposed method with fixed-point arithmetic is 23.1 times faster than the conventional method with floating-point arithmetic. Furthermore, they also show the PSNR of LDR images obtained by the proposed method are comparable to those of the conventional method, though reducing computational and memory cost.
Pavol ZAVARSKY Nobuo FUJII Noriyoshi KAMBAYASHI Masahiro IWAHASHI Somchart CHOKCHAITAM
An unwrapping of signal coefficients in transform domain is proposed for applications in which a lossy operation is performed on the coefficients between analysis and synthesis. It is shown that the unwrapping-based modification of signal-to-additive-signal ratio can employ the fact that an implementation of a biorthogonal decomposition is characterized by a mutually orthogonal eigenvectors. An example to illustrate the benefits of the presented approach in lossy image compression applications is shown.
Xiaoxing ZHANG Xiayu NI Masahiro IWAHASHI Noriyoshi KAMBAYASHI
In this paper, implementation of a first-order active complex filter with variable parameter using operational transconductance amplifiers (OTAs) and grounded copacitors is presented. The proposed configurations can be used as s key building block to realize high-order active complex filters with variable parameter in cascade and leapfrog configuration. Experimental results which are in good agreement with theoretical responses are also given o demonstrate the feasibility of the proposed configurations.
Teerapong ORACHON Taichi YOSHIDA Somchart CHOKCHAITAM Masahiro IWAHASHI Hitoshi KIYA
The lifting wavelet transform (WT) has been widely applied to image coding. Recently, the total number of lifting steps has been minimized introducing a non-separable 2D structure so that delay from input to output can be reduced in parallel processing. However the minimum lifting WT has a problem that its upper bound of the rate-distortion curve is lower than that of the standard lifting WT. This is due to the rounding noise generated inside the transform in its integer implementation. This paper reduces the rounding noise introducing channel scaling. The channel scaling is designed so that the dynamic range of signal values is fully utilized at each channel inside the transform. As a result, the signal to noise ratio is increased and therefore the upper bound of the minimum lifting WT in lossy coding is improved.
Xiaoxing ZHANG Masahiro IWAHASHI Noriyoshi KAMBAYASHI
In this paper a novel narrow-band bandpass filter with an output pair of analytic signals is presented. Since it is based on the complex analog filter, both synthesis and response characteristics of this filter are different from conventional bandpass filters. In the design of this filter, the frequency shift method is employed and the conventional lowpass to bandpass frequency transformation is not required. The analysis and examples show that the output signal pair of the proposed filter possesses same filtering characteristics and a 90 degree phase shifting characteristics in the passband. Therefore, the proposed filter will be used for a single sideband (SSB) signal generator without quadrature generator.