BR102015017870A2 - METHOD FOR COMPUTING THE DCT SPECTRUM FOR ACCUMULATED AND / OR MEDIUM NULL SIGNS - Google Patents

METHOD FOR COMPUTING THE DCT SPECTRUM FOR ACCUMULATED AND / OR MEDIUM NULL SIGNS Download PDF

Info

Publication number
BR102015017870A2
BR102015017870A2 BR102015017870-0A BR102015017870A BR102015017870A2 BR 102015017870 A2 BR102015017870 A2 BR 102015017870A2 BR 102015017870 A BR102015017870 A BR 102015017870A BR 102015017870 A2 BR102015017870 A2 BR 102015017870A2
Authority
BR
Brazil
Prior art keywords
matrix
dct
accumulated
computing
spectrum
Prior art date
Application number
BR102015017870-0A
Other languages
Portuguese (pt)
Inventor
José De Sobral Cintra Renato
Felipe Gomes Coelho Diego
Original Assignee
Universidade Federal De Pernambuco - Ufpe
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Universidade Federal De Pernambuco - Ufpe filed Critical Universidade Federal De Pernambuco - Ufpe
Priority to BR102015017870-0A priority Critical patent/BR102015017870A2/en
Publication of BR102015017870A2 publication Critical patent/BR102015017870A2/en

Links

Landscapes

  • Complex Calculations (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

método para computação do espectro da oct para sinais acumulados e/ou de média nula o método revela um esquema para o cálculo dos coeficientes da dct que pode ser usado em diversos cenários, quer seja em implementação em hardware ou software, e se baseia na fórmula de soma por partes, em que o esquema produzido é capaz de superar métodos convencionais existentes para sinais de entrada em formato comumente encontrado em sistemas de processamento de imagens, reconhecimento de faces e extração de características, e em particular, se torna útil em situações em que o sinal de entrada possui média nula por ser pertinente ao contexto de detecção de característica, por exemplo, em que o nível dc pode não ser relevante, e se torna útil em situações em que o sinal de entrada é nativamente acumulado, e o esquema proposto pode ser útil em problemas de reconhecimento de face, em que os métodos usualmente adotados exigem o dado na forma integrada.Oct Spectrum Computation Method for Accumulated and / or Null Average Signals The method reveals a scheme for calculating dct coefficients that can be used in various scenarios, whether in hardware or software implementation, and is based on the formula summation, where the scheme produced is able to overcome existing conventional methods for format input signals commonly found in image processing, face recognition and feature extraction systems, and in particular, becomes useful in situations in which that the input signal has zero average because it is pertinent to the feature detection context, for example, where the dc level may not be relevant, and becomes useful in situations where the input signal is natively accumulated, and the scheme The proposed approach may be useful in face recognition problems, where the commonly adopted methods require data in the integrated form.

Description

MÉTODO PARA COMPUTAÇÃO DO ESPECTRO DA DCT PARA SINAIS ACUMULADOS E/OU DE MÉDIA NULADCT SPECTRUM COMPUTER METHOD FOR ACCUMULATED AND / OR NULL SIGNALS

[001] A presente invenção trata sobre um esquema de processamento de sinais em que é necessário aplicar a transformada discreta do cosseno de comprimento 8 (DCT). A presente invenção descreve um sistema de processamento de sinais que é capaz de prover o espectro da DCT para o sinal de entrada já integrado, sem a necessidade de bloco de diferenciação. Aplicações de sistemas com essa característica são encontrados em situações em que o sinal seja naturalmente somado, ou que outros passos do sistema completo exija a soma acumulada de um dado sinal intermediário. Isso acontece em aplicações de extração de característica [1], filtragem de imagens integrais [2] e detecção de faces [3]. A invenção se baseia na construção de um esquema de cálculo da DCT que remove a necessidade do bloco de diferenciação.The present invention deals with a signal processing scheme in which it is necessary to apply the discrete length 8 cosine transform (DCT). The present invention describes a signal processing system that is capable of providing the DCT spectrum for the already integrated input signal without the need for differentiation block. Applications of systems with this feature are found in situations where the signal is naturally summed, or where other complete system steps require the accumulated sum of a given intermediate signal. This happens in feature extraction [1], full image filtering [2], and face detection [3] applications. The invention is based on the construction of a DCT calculation scheme that removes the need for the differentiation block.

Antecedente da Invenção [002] A DCT é uma aproximação assintótica para a expansão de Karhunen-Loève, ou como é chamada, a transformada de Karhunen-Loève (KLT), para sinais de Markov estacionários de primeira ordem altamente correlacionados [4], Um típico exemplo de sinais que se encaixam nesta classificação são as imagens [4], Em contextos estatísticos, a KLT é conhecida como análise de componente principal (PCA) [5], com larga aplicação na descorrelação de dados, redução de dimensionalidade e processamento de imagens [6, 7, 8, 9, 10].BACKGROUND OF THE INVENTION DCT is an asymptotic approach to the Karhunen-Loève expansion, or as it is called, the Karhunen-Loève transform (KLT), for highly correlated first-order stationary Markov signals [4]. Typical examples of signals that fit this classification are images [4]. In statistical contexts, KLT is known as principal component analysis (PCA) [5], with wide application in data decorrelation, dimensionality reduction, and data processing. images [6, 7, 8, 9, 10].

[003] A DCT tem sido aplicada em vários contextos práticos: redução de ruído [11], métodos de marca d’água [12], técnicas de compressão de imagem/vídeo [4], e detecção harmônica [4], para citar apenas alguns. Além disso, o recente aumento da demanda de processamento de imagem / vídeo para eletrônicos de consumo [13] e grande manipulação de dados [14] enfatiza a necessidade de computação rápida e eficiente da DCT [15], Como consequência, a DCT é adotada em vários sistemas de imagem e codificação de vídeo [16], tal como JPEG [17], MPEG-1 [17], H.264 [18], e HEVC [19], Com o objetivo de minimizar o custo computacional do cálculo da DCT, uma quantidade de esquemas para o cálculo eficiente da DCT têm sido propostos, incluindo o esquema da DCT de Chen [20], método Lee [21], a arquitetura de Loeffler [22], esquema da DCT-2D de Feig-Winograd DCT [23], e o esquema de Arai da DCT [24].[003] DCT has been applied in a number of practical contexts: noise reduction [11], watermark methods [12], image / video compression techniques [4], and harmonic detection [4] to name a few. just a few. In addition, the recent increased demand for image / video processing for consumer electronics [13] and extensive data manipulation [14] emphasizes the need for fast and efficient computation of DCT [15]. As a consequence, DCT is adopted in various imaging and video encoding systems [16], such as JPEG [17], MPEG-1 [17], H.264 [18], and HEVC [19], in order to minimize the computational cost of calculating A number of schemes for the efficient calculation of DCT have been proposed, including Chen's DCT scheme [20], Lee method [21], Loeffler architecture [22], Feig-DCT-2D scheme. Winograd DCT [23], and the DCT Arai Scheme [24].

[004] De modo geral, um sistema que faz uso do computo do espectro da DCT tem a arqui-tetura apresentada na Figura 1. Contudo, em algumas aplicações [2, 3], o sinal de entrada está acumulado no tempo. Em tais situações, é necessária a introdução do bloco de preprocessamento, em que o sinal acumulado no tempo é submetido a um bloco de recuperação do sinal original. Após o bloco de recuperação do sinal original, este é aplicado ao bloco de que implementa o cálculo da DCT. Este cenário é demonstrado na Figura 2. Em sistemas práticos, qualquer um dos métodos em [20, 21, 22, 23] pode ser usado. Em se tratando de sequências de comprimento 8, os métodos e esquemas em [20, 21, 22] são particularmente úteis.In general, a system that makes use of the DCT spectrum computation has the architecture shown in Figure 1. However, in some applications [2, 3], the input signal is accumulated in time. In such situations, it is necessary to introduce the preprocessing block, wherein the time-accumulated signal is subjected to an original signal retrieval block. After the original signal retrieval block, it is applied to the block from which the DCT calculation is implemented. This scenario is shown in Figure 2. In practical systems, any of the methods in [20, 21, 22, 23] can be used. For length 8 sequences, the methods and schemes in [20, 21, 22] are particularly useful.

[005] A invenção descrita em [27] propõe um método para controle da carga computacional executada com o cálculo dos coeficientes da DCT em codificação de imagens. A invenção descrita em [27] difere da presente invenção por não abordar a forma como a DCT é computada. No trabalho em [27], a DCT pode ser implementada com métodos usuais em [22, 21, 24]. A presente invenção propõe um método para o computo da DCT para sinais com média nula e/ou já acumulados;[005] The invention described in [27] proposes a method for controlling the computational load performed by calculating DCT coefficients in image coding. The invention described in [27] differs from the present invention in that it does not address the way DCT is computed. In working on [27], DCT can be implemented with usual methods in [22, 21, 24]. The present invention proposes a method for computing the DCT for null and / or already accumulated signals;

[006] A invenção descrita em [28] propõe um método para transcodificação de imagens no domínio da DCT. A invenção descrita em [28] difere da presente invenção por não abordar a forma como a DCT é computada. No trabalho em [28], a DCT pode ser implementada com métodos usuais em [22, 21, 24], A presente invenção propõe um método para o computo da DCT para sinais com média nula e/ou já acumulados;[006] The invention described in [28] proposes a method for transcoding images in the DCT domain. The invention described in [28] differs from the present invention in that it does not address the way DCT is computed. In the work in [28], DCT can be implemented with usual methods in [22, 21, 24]. The present invention proposes a method for computing DCT for zero and / or already accumulated signals;

[007] A invenção descrita em [29] propõe um método para controle da carga computacional executada com o cálculo dos coeficientes da DCT em codificação de imagens. A invenção descrita em [29] difere da presente invenção por não abordar a forma como a DCT é computada. No trabalho em [29], a DCT pode ser implementada com métodos usuais em [22, 21, 24], A presente invenção propõe um novo método para o computo da DCT para sinais com média nula e/ou já acumulados;[007] The invention described in [29] proposes a method for controlling the computational load performed by calculating DCT coefficients in image coding. The invention described in [29] differs from the present invention in that it does not address the way DCT is computed. In the work in [29], DCT can be implemented with usual methods in [22, 21, 24]. The present invention proposes a new method for computing DCT for null and / or already accumulated signals;

[008] A invenção descrita em [30] propõe um método de cálculo dos coeficientes da DCT de uma imagem de tamanho arbitrário baseado no cálculo da DCT de blocos locais de tamanhos variáveis e de comprimentos potência de 2. A invenção descrita em [30] difere da presente invenção por não abordar como a DCT dos blocos locais deve ser implementada. No trabalho em [30], a DCT pode ser implementada com métodos usuais em [22, 21, 24], A presente invenção propõe um método para o computo da DCT para sinais com média nula e/ou já acumulados;[008] The invention described in [30] proposes a method of calculating the DCT coefficients of an arbitrary size image based on the DCT calculation of local blocks of variable length and power length of 2. The invention described in [30] differs from the present invention in not addressing how the local block DCT should be implemented. In the work in [30], the DCT can be implemented with usual methods in [22, 21, 24]. The present invention proposes a method for computing the DCT for null and / or already accumulated signals;

[009] A invenção descrita em [31] propõe um método de cálculo dos coeficientes da DCT de blocos de comprimento 8 baseado em blocos de comprimento 4. A invenção descrita em [31] difere da presente invenção por usar a DCT de comprimento 4. No trabalho em [31], a DCT pode ser implementada com métodos usuais [4], A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados;[009] The invention described in [31] proposes a method of calculating the length block 8 DCT coefficients based on the length block 4. The invention described in [31] differs from the present invention by using the length 4 DCT. In the work in [31], DCT can be implemented with usual methods [4]. The present invention proposes a method for computing DCT of length 8 for null and / or already accumulated signals;

[0010] A invenção descrita em [32] propõe um método para a implementação do cálculo dos coeficientes da DCT em microcontroladores baseados no algoritmo de Feig-Winograd [23], A invenção descrita em [32] difere da presente invenção por usar o algoritmo de FeigWinograd [23]. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados;[0010] The invention described in [32] proposes a method for implementing DCT coefficient calculation in microcontrollers based on the Feig-Winograd algorithm [23]. The invention described in [32] differs from the present invention by using the algorithm. from FeigWinograd [23]. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals;

[0011] A invenção descrita em [33] propõe um método para mascaramento espacial de imagens baseado em métricas da imagem de entrada, como variância estimada, gradiente ou coeficientes da DCT. Em particular, quando se trata do uso dos coeficientes da DCT, invenção não especifica o método ou algoritmo rápido usado para o computo dos coeficientes da DCT. A invenção descrita em [33] difere da presente invenção por não especificar o algoritmo para a implementação da DCT A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados;[0011] The invention described in [33] proposes a method for spatial masking of images based on input image metrics such as estimated variance, gradient or DCT coefficients. In particular, when it comes to the use of DCT coefficients, the invention does not specify the fast method or algorithm used for computing DCT coefficients. The invention described in [33] differs from the present invention in that it does not specify the algorithm for the implementation of the DCT. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals;

[0012] A invenção descrita em [34] propõe um método para a extração de dados encriptados em imagens com esteganografia - hidden data in stego-images - após ataque por escalonamento. A invenção descrita em [34] difere da presente invenção pois não trata do computo da DCT. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados;[0012] The invention described in [34] proposes a method for extracting encrypted data in steganography images - hidden data in stego-images - after scaling attack. The invention described in [34] differs from the present invention in that it does not deal with the computation of DCT. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals;

[0013] A invenção descrita em [35] propõe um método para pós-processamento de imagens 3D para redução de ruído. A invenção descrita em [35] difere da presente invenção pois não trata do computo da DCT, mas usa os coeficientes da DCT sem especificar qual método deve ser usado para a computação dos coeficientes. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados.[0013] The invention described in [35] proposes a method for post-processing 3D images for noise reduction. The invention described in [35] differs from the present invention in that it does not deal with the computation of DCT, but uses the coefficients of DCT without specifying which method to use for computing the coefficients. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals.

[0014] A invenção descrita em [36] propõe um método para processamento e codificação de vídeo, baseado no computo dos coeficientes da DCT e em matrizes de quantização modificadas. A invenção descrita em [36] difere da presente invenção pois não trata do computo da DCT, mas usa matrizes de quantização diferentes para aumentar a performance do proposto codificador de video. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados.[0014] The invention described in [36] proposes a method for processing and encoding video based on computation of DCT coefficients and modified quantization matrices. The invention described in [36] differs from the present invention in that it does not deal with the computation of DCT, but uses different quantization arrays to increase the performance of the proposed video encoder. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals.

[0015] A invenção descrita em [37] propõe um método para redução de ruído baseado em segmentação em blocos da imagem original. A invenção descrita em [36] difere da presente invenção pois não trata do computo da DCT, mas apenas faz uso dos coeficientes da DCT para a redução do ruído presente na imagem original. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados;[0015] The invention described in [37] proposes a method for noise reduction based on block segmentation of the original image. The invention described in [36] differs from the present invention in that it does not deal with the computation of DCT, but only makes use of DCT coefficients for noise reduction present in the original image. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals;

[0016] A invenção descrita em [38] propõe um método para a codificação de imagens baseado no computo dos coeficientes da DCT e da transformada discreta wavelet (DWT). A invenção descrita em [38] difere da presente invenção pois não trata do computo da DCT, mas apenas faz uso dos coeficientes da DCT assim como dos coeficientes da DWT. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados;[0016] The invention described in [38] proposes a method for encoding images based on the computation of the DCT and discrete wavelet transform (DWT) coefficients. The invention described in [38] differs from the present invention in that it does not deal with the computation of DCT, but only makes use of DCT coefficients as well as DWT coefficients. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals;

[0017] A invenção descrita em [39] propõe uma técnica para embarcar e detectar marca d’água em imagens. A invenção descrita em [39] difere da presente invenção, pois não trata do computo da DCT, mas apenas faz uso dos coeficientes da DCT para embarcar e detectar a existência de marca d’água. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados;[0017] The invention described in [39] proposes a technique for embedding and detecting watermarks in images. The invention described in [39] differs from the present invention in that it does not deal with the computation of DCT, but only makes use of DCT coefficients to embark and detect the existence of watermark. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals;

[0018] A invenção descrita em [40] propõe uma técnica para embarcar marca d’água em sinais de imagem e áudio baseada em blocos. A invenção descrita em [40] difere da presente invenção, pois não trata do computo da DCT, mas apenas faz uso dos coeficientes da DCT para embarcar marca d água na imagem de entrada. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com media nula e/ou já acumulados;[0018] The invention described in [40] proposes a technique for embedding watermark on block-based image and audio signals. The invention described in [40] differs from the present invention in that it does not deal with the DCT computation, but only makes use of the DCT coefficients to embed watermark in the input image. The present invention proposes a method for computing DCT of length 8 for zero and / or accumulated media signals;

[0019] A invenção descrita em [41] propõe uma técnica de redução de ruído em imagens baseada em um sistema variante do esquema de filtragem de Wiener com imagens em representação piramidal. A invenção descrita em [41] difere da presente invenção pois não trata do computo da DCT, mas apenas faz uso dos coeficientes da DCT a redução do ruído presente na imagem original. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados;[0019] The invention described in [41] proposes an image noise reduction technique based on a variant system of the Wiener filtering scheme with pyramidal images. The invention described in [41] differs from the present invention in that it does not deal with the computation of DCT, but only makes use of DCT coefficients to reduce the noise present in the original image. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals;

[0020] A invenção descrita em [42] propõe um método para processamento de imagens baseado em decimação e interpolação da imagem original. A invenção descrita em [42] difere da presente invenção pois não trata do computo da DCT, mas apenas faz uso dos coeficientes da DCT baseado na separação da parte par e da parte ímpar. A presente invenção propõe um método para o computo da DCT de comprimento 8 para sinais com média nula e/ou já acumulados;[0020] The invention described in [42] proposes a method for image processing based on decimation and interpolation of the original image. The invention described in [42] differs from the present invention in that it does not deal with the computation of the DCT, but only makes use of the DCT coefficients based on the separation of the even part and the odd part. The present invention proposes a method for computing the DCT of length 8 for zero average and / or already accumulated signals;

Sumário da Invenção [0021] A invenção, descrita neste relatório, revela um esquema para o cálculo dos coeficientes da DCT que pode ser usado em diversos cenários, quer seja em implementação em hardware ou software Este esquema se baseia na fórmula de soma por partes. O esquema produzido é capaz de superar métodos convencionais existentes [20, 21, 22, 24] para sinais de entrada em formato comumente encontrado em sistemas de processamento de imagens [2], reconhecimento de faces [3] e extração de características [1], Em particular, a presente invenção se torna útil em situações em que o sinal de entrada possui média nula ou já é nativamente acumulado. Situações em que o sinal apresenta média nula é pertinente ao contexto de detecção de característica, por exemplo, em que o nível DC pode não ser relevante [25, 1], Em situações em que o sinal de entrada é nativamente acumulado, o esquema proposto pode ser útil em problemas de reconhecimento de face, em que os métodos usualmente adotados exigem o dado na forma integrada [2, 3], Descrição das Figuras [0022] A Figura 1 apresenta um sistema genérico usual que computa a DCT de uma dada sequência arbitrária de entrada x [n];Summary of the Invention The invention, described in this report, discloses a scheme for calculating DCT coefficients that can be used in various scenarios, whether in hardware or software implementation. This scheme is based on the piecewise sum formula. The scheme produced is capable of overcoming existing conventional methods [20, 21, 22, 24] for format input signals commonly found in image processing systems [2], face recognition [3] and feature extraction [1] In particular, the present invention becomes useful in situations where the input signal has zero average or is already natively accumulated. Situations where the signal has zero mean is pertinent to the characteristic detection context, for example, where the DC level may not be relevant [25, 1]. In situations where the input signal is natively accumulated, the proposed scheme may be useful in face recognition problems where commonly adopted methods require data in the integrated form [2, 3]. Description of the Figures [0022] Figure 1 presents a usual generic system that computes the DCT of a given sequence. arbitrary input x [n];

[0023] A Figura 2 apresenta um sistema genérico usual que computa a DCT de uma dada sequência arbitrária de entrada x [n] quando é dada como entrada a sequência na forma acumulada z [n];[0023] Figure 2 shows a usual generic system that computes the DCT of a given arbitrary input sequence x [n] when the sequence is accumulated as input z [n];

[0024] A Figura 3 apresenta a arquitetura do sistema proposto para o cenário (i) quando o sinal de entrada possui média nula, denotado por x~[n], A sequência z-[n] é a sequência acumulada construída a partir de x~[n], O bloco de acumulação pode ser implementado de modo usual com técnicas convencionais [26];Figure 3 presents the system architecture proposed for scenario (i) when the input signal has zero mean, denoted by x ~ [n]. The sequence z- [n] is the accumulated sequence constructed from x ~ [n], The accumulation block can be implemented in usual manner with conventional techniques [26];

[0025] A Figura 4 apresenta a arquitetura do sistema proposto para o cenário (ii) quando o sinal de entrada é naturalmente acumulado. A sequência z~[n] é a sequência acumulada construída a partir de x-[n];Figure 4 presents the proposed system architecture for scenario (ii) when the input signal is naturally accumulated. The sequence z ~ [n] is the accumulated sequence constructed from x- [n];

[0026] A Figura 5 apresenta o circuito esquemático proposto para o cálculo dos coeficientes da DCT escalonados—exceto pela ausência do termo DC de x [n];[0026] Figure 5 presents the proposed schematic circuit for calculating staggered DCT coefficients — except for the absence of the DC term of x [n];

[0027] A Figura 6 apresenta o circuito esquemático proposto para o cálculo dos coeficientes da DCT exata—exceto pela ausência do termo DC de x [n];[627] Figure 6 presents the proposed schematic circuit for calculating the exact DCT coefficients — except for the absence of the DC term of x [n];

[0028] A Figura 7 apresenta o circuito equemático proposto para a remoção do nível DC de x [n] da sequência acumulada z [n], resultando em z~[n].[0028] Figure 7 presents the proposed equemic circuit for removing the DC level of x [n] from the accumulated sequence z [n], resulting in z ~ [n].

Descrição Detalhada [0029] A presente invenção difere em aspectos relevantes de sistemas e esquemas já propostos em trabalhos relacionados.Detailed Description The present invention differs in relevant aspects from systems and schemes already proposed in related works.

[0030] A técnica de soma por partes é o equivalente em tempo discreto ao método integração por partes [43], Seja x [n] e y [n] dois sinais de tempo discreto. A soma por partes prescreve que [43, 44]: Ay[n] â y\n + 1] — y\n] |33]. em que Δ denota o operador de diferença finita dado por [0031] A expressão acima pode ser simplificada com a suposição das seguintes condições fracas adicionais. Admitindo que os sinais são considerados periódicos com período N, foi estabelecido, em [45] que: [0032] A condição acima sobre a periodicidade não é demasiadamente restritiva. A teoria de transformadas discretas assume que os sinais de entrada são periódicos [46, 4, 47], Esse é o caso em todas as transformações listadas na Tabela 1. Em particular, a DCT pode ser obtida como a solução de um problema de oscilação harmônica [4], [0033] A expressão pode ser interpretada como uma transformação de tempo discreto. Seja x [η] o sinal de entrada a ser transformado e y[n] = ker[n, k] o núcleo de uma transformação para k-ésima componente no domínio da transformada. Portanto, temos que: em que X[k] é o sinal que representa as componentes da transformada. A Tabela 1 apresenta alguns núcleos de transformações usuais. Núcleos de transformações usuais [0034] Portanto, aplicando (1) em (2) produz a seguinte expressão para as components no domínio espectral: em que para n = 0, 1....................n - 1.[0030] The piecewise sum technique is the discrete time equivalent of the piecewise integration method [43], let x [n] and y [n] be two discrete time signals. The sum by parts prescribes that [43, 44]: Ay [n] â y \ n + 1] - y \ n] | 33]. where Δ denotes the finite difference operator given by [0031] The above expression can be simplified by assuming the following additional weak conditions. Assuming that signals are considered periodic with period N, it has been established in [45] that: [0032] The above condition on periodicity is not too restrictive. The discrete transform theory assumes that the input signals are periodic [46, 4, 47]. This is the case in all the transformations listed in Table 1. In particular, DCT can be obtained as the solution of an oscillation problem. harmonic [4], [0033] The expression can be interpreted as a discrete time transformation. Let x [η] be the input signal to be transformed and y [n] = ker [n, k] the nucleus of a transformation for kth component in the transform domain. Therefore, we have that: where X [k] is the signal that represents the components of the transform. Table 1 presents some usual transformation nuclei. Usual Transformation Cores Therefore, applying (1) to (2) produces the following expression for the components in the spectral domain: where for n = 0, 1 .............. ...... n - 1.

[0035] Comparando (2) com (3), percebe-se que a expressão original da transformação foi reescrita em uma forma alternativa em que ambos os dados de entrada e o núcleo foram processados. Observe que z [n] é a saída de um sistema acumulador para o sinal de entrada x [n] [36]; enquanto Δ ker[n, k] deriva de um sistema de diferenças finitas para o sinal de entrada ker[n, k] [46]. Embora o sistema de diferenças finitas não seja causai, este fato não representa nenhuma dificuldade ao formalismo acima. Isto acontece porque ker[n, k] não é uma sequência aleatória — mas uma sequência determinística cujos valores são conhecidos a priori [48, p.7], [0036] Observa-se que se x [n] possue média nula, o seguinte é válido: [0037] Para transformações trigonométricas, condição acima implica X[0] = 0 (valor DC nulo). Portanto, (3) pode ser simplificada e escrito como: [0038] 0 somatório acima varia de 0 a N - 2. Isto significa que a matriz da transformação associada a (3) tem dimensão (N - 1) * (N - 1). Isto contrasta com a matriz de transformação original, que tem tamanho N * n. Assim, a soma por partes provocou uma redução de dimensão na computação da transformada. Como consequência, o custo computacional de algoritmos associados espera-se ser reduzido.Comparing (2) with (3), it is clear that the original expression of the transformation was rewritten in an alternative way in which both input data and the core were processed. Note that z [n] is the output of an accumulator system for input signal x [n] [36]; whereas Δ ker [n, k] derives from a finite difference system for the input signal ker [n, k] [46]. Although the finite difference system is not causal, this fact poses no difficulty to the above formalism. This is because ker [n, k] is not a random sequence - but a deterministic sequence whose values are known a priori [48, p.7], [0036] Note that if x [n] has zero mean, the following is valid: [0037] For trigonometric transformations, condition above implies X [0] = 0 (null DC value). Therefore, (3) can be simplified and written as: The sum above ranges from 0 to N - 2. This means that the transformation matrix associated with (3) has dimension (N - 1) * (N - 1 ). This contrasts with the original transformation matrix, which has size N * n. Thus, the piecewise sum has caused a reduction in dimension in the computation of the transform. As a consequence, the computational cost of associated algorithms is expected to be reduced.

[0039] A DCT é uma transformação linear que mapeia um sinal de tempo discreto x [n] de comprimento N em ou outro sinal de tempo discreto X[k] de comprimento N, de acordo com a seguinte relação [22]: em que ak = 1/V2, para k = 0 e ak =1 [0040] Caso contrário, a expressão acima pode ser apresentada em forma compacta por meio de representação matricial.DCT is a linear transformation that maps a discrete time signal x [n] of length N to another discrete time signal X [k] of length N according to the following relationship [22]: where ak = 1 / V2, for k = 0 and ak = 1 Otherwise, the above expression may be presented in compact form by matrix representation.

[0041] Considerando sinais x [n] e X[k] em formato vetorial como temos que: em que CNéa matriz da DCT, cuja entrada (k, n) é dada por Para N - 8, temos a seguinte matriz de transformação: em que Ck = cos(kTT/16), para k = 1, 2...7.Considering x [n] and X [k] signals in vector format as follows: where CN is the DCT matrix, whose input (k, n) is given by For N - 8, we have the following transformation matrix: where Ck = cos (kTT / 16), for k = 1,2, ... 7.

[0042] Esta definição particular da DCT é adotada em [39, 40, 41], tendo sido considerado por derivar o conhecido esquema de Loeffler [22], Observa-se que V2 · c4 = 1; portanto, a componente DC, X[0], é avaliada sem multiplicações [22], [0043] Para facilitar o desenvolvimento do esquema proposto, foi apresentada a fórmula da soma por partes no formalismo matricial. Em primeiro lugar, o operador de diferenças finitas precisa ser adaptado para manipular matrizes. Seja M uma matriz quadrada. Então, seja ΔΜ a matriz resultante da aplicação do operador de diferenças finitas usual em cada uma das linhas de M.[0042] This particular definition of DCT is adopted in [39, 40, 41], having been considered to derive the well-known Loeffler scheme [22]. It is noted that V2 · c4 = 1; therefore, the DC component, X [0], is evaluated without multiplication [22], [0043] To facilitate the development of the proposed scheme, the piecewise sum formula in matrix formalism was presented. First, the finite difference operator must be adapted to manipulate arrays. Let M be a square matrix. Then let ΔΜ be the matrix resulting from the application of the usual finite difference operator on each of the M lines.

[0044] Considerando-se a fórmula de soma por partes, mostrada em (4), tem-se que (5) pode ser escrito de acordo com: X = AC · z, em que o vetor epresenta o espectro da DCT, é o sinal acumulado e [0045] Considerando as identidades de soma-produto [52, p. 72] e as identidades de simetrias [43], as entradas da matriz AC podem ser dadas em forma multiplicativa, como mostrado abaixo: para sk = sin(kTT/16), k = 1, 2..7.Considering the piecewise sum formula shown in (4), one has that (5) can be written according to: X = AC · z, where the vector represents the DCT spectrum, is the accumulated sign and [0045] Considering the sum-product identities [52, p. 72] and symmetry identities [43], the AC matrix inputs can be given in multiplicative form, as shown below: for sk = sin (kTT / 16), k = 1, 2..7.

[0046] Visto que o sinal de entrada x [n] é assumido por ter média nula, temos que X[0] = 0 e z [N -1] = 0. Consequentemente, a primeira linha e última coluna de AC pode ser desprezada. Isso implica que apenas a submatriz de dimensão 7 * 7 é necessária para a computação de X. Esta particular matriz é dada por: [0047] Note que C é suficiente para a computação de todos os coeficientes da DCT — exceto o nível DC.Since the input signal x [n] is assumed to have zero mean, we have that X [0] = 0 and z [N -1] = 0. Consequently, the first row and last column of AC can be neglected. . This implies that only the 7 * 7 dimension subarray is required for X computation. This particular matrix is given by: [0047] Note that C is sufficient for computation of all DCT coefficients - except the DC level.

[0048] Um exame da matriz C revela a existência de multiplicadores repetidos ao longo das linhas. Assim, a seguinte fatoração é obtida: em que S = 2V2 · diag(s1, s2, s3, s4, s5, s6, s7). A expressão acima nos diz que a matriz S contribui apenas com fatores de escalonamentos para a computação dos coeficientes da DCT. Considerando aplicações em que é necessário apenas uma versão escalonada do espectro da DCT—como detecção harmônica [53] e melhoramento de cores [54]—o custo computacional de S pode ser desconsiderado. Adicionalmente, em contextos de compressão de imagens, matrizes diagonais podem ser absorvidas diretamente na matriz de quantização; não representando computação extra [55, 56], [0049] Note que o fator de escalonamento 2V2S4 = 2 é uma multiplicação trivial [26] que pode ser implementada a partir de um simples deslocamento de bits. Portanto, o custo computacional da matriz de escalonamento S é de apenas seis—não sete—multiplicações. O objetivo é prover uma fatoração por meio de matrizes esparsas para C. Visto que C apresenta grande quantidade de simetrias, métodos de fatoração baseados em estruturas borboletas podem ser diretamente aplicados [46, 4, 26]. Portanto, obtêm-se a seguinte fatoração: C = S P Μ A. em que e [0050] A matriz P é uma simples matriz de permutação, que não representa custo computacional. Em termos de hardware, P representa uma modificação nas ligações internas. A matriz A é uma matriz aditiva que consiste que um estágio borboleta usual presente em algoritmos de decimação em frequência [26], A matriz restante M é bloco diagonal e ainda contém redundâncias matemáticas devido sua natureza simétrica. Considerando a fatoração matricial para projeto de algoritmos rápidos [26], a seguinte expressão pode ser obtida: Μ = M1 · M2 · M3 · M4, em que [0051] A matriz bloco diagonal M2 contém um bloco de rotação [4] que pode ser decomposto. Então, temos: M2 = R1 · R2 · R3, em que e R2 = diag(1,1,82,1,1,1,1,1).An examination of matrix C reveals the existence of repeated multipliers along the lines. Thus, the following factorization is obtained: where S = 2V2 · diag (s1, s2, s3, s4, s5, s6, s7). The above expression tells us that the S matrix only contributes scaling factors to compute the DCT coefficients. Considering applications where only a scaled version of the DCT spectrum is required — such as harmonic detection [53] and color enhancement [54] —the computational cost of S can be disregarded. Additionally, in image compression contexts, diagonal matrices can be absorbed directly into the quantization matrix; not representing extra computation [55, 56], [0049] Note that the scaling factor 2V2S4 = 2 is a trivial multiplication [26] that can be implemented from a simple bit shift. Therefore, the computational cost of the scaling matrix S is only six — not seven — multiplications. The objective is to provide factorization through sparse matrices for C. Since C has a large amount of symmetries, factoring methods based on butterfly structures can be directly applied [46, 4, 26]. Therefore, the following factorization is obtained: C = S P Μ A. where e [0050] The matrix P is a simple permutation matrix, which does not represent computational cost. In hardware terms, P represents a change in internal connections. Matrix A is an additive matrix consisting of a usual butterfly stage present in frequency decimation algorithms [26]. The remaining matrix M is diagonal block and still contains mathematical redundancies due to its symmetrical nature. Considering the matrix factorization for fast algorithm design [26], the following expression can be obtained: Μ = M1 · M2 · M3 · M4, where [0051] The diagonal block matrix M2 contains a rotation block [4] that can be decomposed. Then we have: M2 = R1 · R2 · R3, where R2 = diag (1,1,82,1,1,1,1,1).

[0052] Por meio de manipulações trigonométricas usuais, os multiplicandos requeridos em Rt satisfazem: S2 + S6 = V2 C2 e S6 - S2 = V2 S2. Finalmente, a fatoração de matrizes esparsas é dada por: C~ = S · P · M1 · R1 · R2 · R3 · M3 · M4 · A.Through usual trigonometric manipulations, the required multiplicands in Rt satisfy: S2 + S6 = V2 C2 and S6 - S2 = V2 S2. Finally, the factorization of sparse matrices is given by: C ~ = S · P · M1 · R1 · R2 · R3 · M3 · M4 · A.

[0053] O ciruito esquemático resultante para o proposto método é apresentado na Figura 6. O circuito esquemático resultante para o proposto método com o computo do espectro da DCT escalonado é apresentado na Figura 5.The resulting schematic circuit for the proposed method is shown in Figure 6. The resulting schematic circuit for the proposed method with the scaled DCT spectrum computation is shown in Figure 5.

[0054] No que segue, a complexidade computacional é avaliada, o que é consequência direta da proposição dos circuitos esquemáticos propostos para diversos cenários. A complexidade computacional é avaliada em termos do número de somas e multiplicações não triviais requerida para a computação da DCT por meio do proposto método.In what follows, computational complexity is evaluated, which is a direct consequence of the proposition of the proposed schematic circuits for various scenarios. Computational complexity is evaluated in terms of the number of nontrivial sums and multiplications required for the computation of DCT by the proposed method.

[0055] São apresentados os diferentes cenários para a forma do sinal de entrada: (i) sinal de média nula; (ii) sinal acumulado; e (iii) sinal acumulado e de média nula. O cenário (i) é pertinente ao contexto de detecção de característica, por exemplo, onde o nível DC pode não ser relevante [25, 1]. O cenário (ii) representa o caso em que o sinal de entrada é nativamente acumulado (integrado). Isso pode acontecer em problemas de reconhecimento de face, em que os algoritmos usualmente adotados exigem o dado na forma integrada [2, 3]. O cenário (iii) é uma combinação dos dois cenários acima.The different scenarios for the shape of the input signal are presented: (i) null average signal; (ii) accumulated signal; and (iii) zero average accumulated signal. Scenario (i) is relevant to the feature detection context, for example, where the DC level may not be relevant [25, 1]. Scenario (ii) represents the case where the input signal is natively accumulated (integrated). This can happen in face recognition problems, in which the usually adopted algorithms require the data in the integrated form [2, 3]. Scenario (iii) is a combination of the two scenarios above.

[0056] O bloco de remoção do nível DC exige 6 subtrações para retirar a média do sinal x [n] de cada uma das amostras z [0], z [1]...... z [6]. O custo computacional decorrente de multiplicações são concentrados nas matrizes S, Ri, R2, e M3. Estas exigem um total de 11 multiplicações não triviais, que é a complexidade computacional mínima teórica para a DCT de comprimento 8 [57], A matrizes S, P, M3, e R2 contribuem com 19 adições.The DC level removal block requires 6 subtractions to average the signal x [n] from each of the samples z [0], z [1] ...... z [6]. The computational cost resulting from multiplications is concentrated in matrices S, Ri, R2, and M3. These require a total of 11 nontrivial multiplications, which is the theoretical minimum computational complexity for the DCT of length 8 [57]. The matrices S, P, M3, and R2 contribute 19 additions.

[0057] A Tabela 2 compara a complexidade computacional do circuito esquemático proposto com diversas outras técnicas largamente adotadas para a computação da DCT nos cenários destacados. Para cada cenário, enfatizamos os melhores resultados em negrito. Visto que o método de Arai [24] e o método proposto são métodos capazes de prover o espectro escalonado da DCT, mostramos a complexidade do método escalonado também. Em parênteses é apresentado a complexidade multiplicativa completa quando os fatores de escalonamento sâo considerados.[0057] Table 2 compares the computational complexity of the proposed schematic circuit with several other widely adopted techniques for computing DCT in the highlighted scenarios. For each scenario, we emphasize the best results in bold. Since the Arai method [24] and the proposed method are methods capable of providing the scaled spectrum of the DCT, we show the complexity of the scaled method as well. In parentheses the complete multiplicative complexity is presented when the scaling factors are considered.

[0058] Comparação das multiplicações não triviais e adições para os métodos para a computação da DCTComparison of Nontrivial Multiplications and Additions to Methods for Computing DCT

[0059] O método proposto, neste relatório, superou todos os métodos concorrentes nos cenários (i), (ii), e (iii).The method proposed in this report outperformed all concurrent methods in scenarios (i), (ii), and (iii).

[0060] Embora o método proposto exija cinco multiplicações para computar a DCT escalonada, o mesmo que Arai, o método proposto demanda uma menor quantidade de multiplicações.Although the proposed method requires five multiplications to compute the staggered DCT, the same as Arai, the proposed method requires a smaller number of multiplications.

Bibliografia [0061] [1] Y. Wang, C.-S. Chua, and Y.-K. Ho, “Facial feature detection and face recognition from 2D and 3D images," Pattern Recognition Letters, vol. 23, no. 10, pp. 1191-1202, 2002.Bibliography [0061] [1] Y. Wang, C.-S. Chua, and Y.-K. Ho, “Facial feature detection and face recognition from 2D and 3D images,” Pattern Recognition Letters, vol. 23, no. 10, pp. 1191-1202, 2002.

[0062] [2] Elboher and M. Werman, “Efficient and accurate Gaussian image filtering using running sums," in 12th International Conference on Intelligent Systems Design and Applications (ISDA), 2012, pp. 897-902.[0062] [2] Elboher and M. Werman, “Efficient and accurate Gaussian image filtering using running sums,” in 12th International Conference on Intelligent Systems Design and Applications (ISDA), 2012, pp. 897-902.

[0063] [3] P. Viola and M. Jones, “Rapid object detection using a boosted cascade of simpl features," in IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 1, 2001.[0063] [3] P. Viola and M. Jones, "Rapid Object Detection Using a Boosted Cascade of Simple Features," in IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 1, 2001.

[0064] [4] V. Britanak, P. Yip, and K. R. Rao, Discrete Cosine and Sine Transforms.Academic Press, 2007.V. Britanak, P. Yip, and K. R. Rao, Discrete Cosine and Sine Transforms. Academic Press, 2007.

[0065] [5] C. M. Sousa Junior, “Compressão de imagens utilizando análise de componentes independentes,” Master's thesis, Universidade Federal do Maranhão, São Luís-MA, Brazil,2007.[0065] [5] C. M. Sousa Junior, “Image Compression Using Independent Component Analysis,” Master's thesis, Federal University of Maranhão, São Luís-MA, Brazil, 2007.

[0066] [6] X. Zou, P. Jancovic, and M. Kokuer, “On the effectiveness of the ICA-based signal representation in non-Gaussian noise,” in 9th International Conference on Signal Processing,2008, pp. 1-4.[0066] [6] X. Zou, P. Jancovic, and M. Kokuer, “On the Effectiveness of the ICA-based Signal Representation in Non-Gaussian Noise,” in 9th International Conference on Signal Processing, 2008, p. 1-4.

[0067] [7] Kolossa, H. Sawada, R. F. Astudillo, R. Orglmeister, and S. Makino, “Recognition of convolutive speech mixtures by missing feature techniques for ICA,” in Fortieth Asilomar Conference on Signals, Systems and Computers, 2006, pp. 1397-1401.[0067] [7] Kolossa, H. Sawada, RF Astudillo, R. Orglmeister, and S. Makino, “Recognition of Convolutive Speech Mixtures by Missing Feature Techniques for ICA,” in Fortieth Asilomar Conference on Signals, Systems and Computers, 2006 , pp. 1397-1401.

[0068] [8] Y. Zhang and W. H. Abdulla, “Eigenanalysis applied to speaker Identification using gammatone auditory filterbank and independent component analysis,” in 9th International Symposium on Signal Processing and Its Applications (ISSPA), 2007, pp. 1-4.[0068] [8] Y. Zhang and W. H. Abdulla, “Eigenanalysis applied to speaker Identification using gammatone auditory filterbank and independent component analysis,” in 9th International Symposium on Signal Processing and Its Applications (ISSPA), 2007, pp. 1-4.

[0069] [9] Balakrishna, P. V. A. Kumar, C. Prakash, and S. V. Gangashetty, “Speech enhancement using ICA with Bessel features," in 18th International Conference on Systems, Signals and Image Processing (IWSSIP), 2011, pp. 1-4.[0069] [9] Balakrishna, Kumar PVA, C. Prakash, and SV Gangashetty, “Speech enhancement using ICA with Bessel features," in 18th International Conference on Systems, Signals and Image Processing (IWSSIP), 2011, pp. 1- 4

[0070] [10] T. J.-H. Lee, H.-Y. Jung, T.-W. Lee, and S.-Y. Lee, “Speech feature extraction using independent component analysis," in IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), vol. 3, 2000, pp. 1631-1634.[10] T. J.-H. Lee, H.-Y. Jung, T.-W. Lee, and S.-Y. Lee, “Speech feature extraction using independent component analysis,” in IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), vol. 3, 2000, pp. 1631-1634.

[0071] [11] S. K. Gupta, J. Jain, and R. Pachauri, “Improved noise cancellation in discrete cosine transform domain using adaptive block LMS filter," International Journal of Engineering Science and Advanced Technology, vol. 2, no. 3, pp. 498-502, Jun. 2012.[0071] [11] SK Gupta, J. Jain, and R. Pachauri, “Improved noise cancellation in discrete cosine transform domain using adaptive block LMS filter,” International Journal of Engineering Science and Advanced Technology, Vol. 2, No. 3 , pp. 498-502, Jun. 2012.

[0072] [12] S. An and C. Wang, “A computation structure for 2-D DCT watermarking,” in 52nd IEEE International Midwest Symposium on Circuits and Systems, Aug. 2009, pp. 577-580.[0072] [12] S. An and C. Wang, “A computation structure for 2-D DCT watermarking,” in 52nd IEEE International Midwest Symposium on Circuits and Systems, Aug. 2009, pp. 577-580.

[0073] [13] A. A. Chien and V. Karamcheti, “Moore’s law: The first ending and a new beginning," Computer, vol. 46, no. 12, pp. 48-53, 2013.[0073] [13] A. A. Chien and V. Karamcheti, "Moore's Law: The First Ending and a New Beginning," Computer, vol. 46, no. 12, pp. 48-53, 2013.

[0074] [14] M. R. Wigan and R. Clarke, “Big data’s big unintended consequences," Computer, vol. 46, no. 6, pp. 46-53, 2013.[0074] [14] M. R. Wigan and R. Clarke, “Big Data's Big Unintended Consequences,” Computer, vol. 46, no. 6, pp. 46-53, 2013.

[0075] [15] L. Ji and L. X. Ming, "New DCT computation algorithm for VLIW architecture,” in 6,h International Conference on Signal Processing, vol. 1, 2002, pp. 41-44.[0075] [15] L. Ji and L. X. Ming, "New DCT Computation Algorithm for VLIW architecture," in 6, International Conference on Signal Processing, vol. 1, 2002, pp. 41-44.

[0076] [16] V. Bhaskaran and K. Konstantinides, Image and Video Compression Standards. Kluwer Academic Publishers, Jun. 1997.[0076] [16] V. Bhaskaran and K. Konstantinides, Image and Video Compression Standards. Kluwer Academic Publishers, Jun. 1997.

[0077] [17] G. K. Wallace, The JPEG still picture compression standard,” IEEE Transactions on Consumer Electronics, vol. 38, no. 1, pp. xviii-xxxiv, Feb 1992.[0077] [17] G. K. Wallace, The JPEG Still Picture Compression Standard, ”IEEE Transactions on Consumer Electronics, Vol. 38, no. 1, pp. xviii-xxxiv, Feb 1992.

[0078] [18] T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “OverView of the H.264/AVC video coding standard," IEEE Transactions on Circuit and Systems for Video Technology, vol. 13, no. 7, pp. 560-576, Jul. 2003.T. Wiegand, GJ Sullivan, G. Bjontegaard, and A. Luthra, "OverView of the H.264 / AVC Standard Video Coding," IEEE Transactions on Circuit and Systems for Video Technology, vol. 13, No. 7, pp. 560-576, Jul. 2003.

[0079] [19] Μ. T. Pourazad, C. Doutre, M. Azimi, and P. Nasiopouios, “HEVC: The new gold standard for video compression: How does HEVC compare with H.264/AVC?” IEEE Consumer Electronics Magazine, vol. 1, no. 3, pp. 36^46, Jul. 2012.[0079] [19] Μ. T. Pourazad, C. Doutre, M. Azimi, and P. Nasiopouios, “HEVC: The New Gold Standard for Video Compression: How Does HEVC Compare with H.264 / AVC?” IEEE Consumer Electronics Magazine, Vol. 1, no. 3, pp. 36-46, Jul. 2012.

[0080] [20] H. W.-H. Chen, C. Smith, and S. Fralick, “A fast computational algorithm for the discrete cosine transform,” IEEE Transactions on Communications, vol. 25, no. 9, pp. 1004-1009, Jan. 1977.[20] H. W.-H. Chen, C. Smith, and S. Fralick, “A fast computational algorithm for the discrete cosine transform,” IEEE Transactions on Communications, vol. 25, no. 9, pp. 1004-1009, Jan. 1977.

[0081] [21] B. Lee, “A new algorithm to compute the discrete cosine transform," IEEE Transactions on Acoustics, Speech and Signal Processing, vol. 32, no. 6, pp. 1243-1245, Dec. 1984.[0081] [21] B. Lee, “A new algorithm for computing the discrete cosine transform," IEEE Transactions on Acoustics, Speech and Signal Processing, vol. 32, no. 6, pp. 1243-1245, Dec. 1984.

[0082] [22] C. Loeffler, A. Ligtenberg, and G. S. Moschytz, “A practical fast 1-D DCT algorithms with 11 multiplications,” in International Conference on Acoustics, Speech, and Signal Processing, vol. 2, May 1989, pp. 988-991.C. Loeffler, A. Ligtenberg, and G.S. Moschytz, “A Practical Fast 1-D DCT Algorithms with 11 Multiplications,” in International Conference on Acoustics, Speech, and Signal Processing, vol. 2, May 1989, pp. 988-991.

[0083] [23] E. Feig and S. Winograd. “Fast algorithms for the discrete cosine transform,” IEEE Transactions on Signal Processing, vol. 40, no. 9, pp. 2174-2193, Sep. 1992.[0083] [23] E. Feig and S. Winograd. “Fast algorithms for the discrete cosine transform,” IEEE Transactions on Signal Processing, Vol. 40, no. 9, pp. 2174-2193, Sep. 1992

[0084] [24] Y. Arai, Τ. Agui, and M. Nakajima, “A fast DCT-SQ scheme for images," IEICE Transactions, vol. E71, no. 11, pp. 1095-1097, Nov. 1988.[24] Y. Arai, Τ. Agui, and M. Nakajima, "A fast DCT-SQ scheme for images," IEICE Transactions, vol. E71, no. 11, pp. 1095-1097, Nov. 1988.

[0085] [25] A. K. Jain, Fundamentais of Digital Image Processing, T. Kailath, Ed. University of Califórnia, Davis: Prentice Hall Information and System Sciences Series, 1989.[0085] [25] A. K. Jain, Fundamentals of Digital Image Processing, T. Kailath, Ed. University of California, Davis: Prentice Hall Information and System Sciences Series, 1989.

[0086] [26] R. E. Blahut, Fast Algorithms for Digital Signal Processing. Cambridge University Press, 2010.[0086] [26] R. E. Blahut, Fast Algorithms for Digital Signal Processing. Cambridge University Press, 2010.

[0087] [27] K. W. Chun and B. C. Song, “Method and apparatus for controlling amount of DCT computation performed to encode motion image,’’ Patent EP20 030 012 433, Jan. 14 , 2004, eP Patent App. EP20,030,012,433. [Online]. Available: [0088] [28] C. W. Lin, Η. K. Tsai, and J. J. Luo, "Methods and Systems for video transcoding in DCT domain with low complexity," Patent US7 236 529, Jun. 26, 2007, uS Patent 7,236,529. [Online], Available: [0089] [29] B. Song and K. Chun, “Method and apparatus for controlling amount of DCT computation performed to encode motion image," Patent US10/600 654, Jan. 15, 2004, uS Patent App. 10/600,654. [Online], Available: [0090] [30] K. Jongil, “Fast efficient computation of the discrete cosine transform (dct) of digital video data," Patent US2001/040 999, Jan. 10, 2002, wO Patent App. PCT/US2001/040,999. [Online], Available: [0091] [31] F. Pan, “Method and system for computing 8x8 dct/idct and a vlsi implementation," Patent US1999/002 186, Aug. 5, 1999, wO Patent App. PCT/US1999/002,186. [Online], Available: [0092] [32] L. Cali’ and P.L. Rolandi, “Method and Circuit for computing the discrete cosine transform (DCT) in microcontrollers,” Patent US6 493 737, Dec. 10, 2002, uS Patent 6,493,737.[0087] [27] K.W. Chun and B.C. Song, 'Method and Apparatus for Controlling the Amount of DCT Computation Performed to Encode Motion Image,' 'Patent EP20 030 012 433, Jan. 14, 2004, and Patent Application. EP20,030,012,433. [Online]. Available: [0088] [28] C. W. Lin, Η. K. Tsai, and J.J. Luo, "Methods and Systems for Video Transcoding in DCT Domain with Low Complexity," Patent US 7,236,529, Jun. 26, 2007, US Patent 7,236,529. [Online], Available: [0089] [29] B. Song and K. Chun, “Method and apparatus for controlling the amount of DCT computation performed to encode motion image,” Patent US10 / 600 654, Jan. 15, 2004, uS Patent App. 10 / 600,654. [Online], Available: [0090] [30] K. Jongil, “Fast efficient computation of the discrete cosine transform (dct) of digital video data,” Patent US2001 / 040 999, Jan. 10 , 2002, Patent Application. PCT / US2001 / 040,999. [Online], Available: [0091] [31] F. Pan, “Method and system for computing 8x8 dct / idct and a vlsi implementation,” Patent US1999 / 002 186, Aug. 5, 1999, The Patent App. PCT / US1999 / 002,186. [Online], Available: [0092] [32] L. Cali 'and PL Rolandi, "Method and Circuit for Computing the Discrete Cosine Transform (DCT) in Microcontrollers," Patent US6 493 737, Dec. 10, 2002, US Patent 6,493,737.

[0093] [33] X. Lu, J. Zhai, and C. Gomila, “Spatial masking using a spatial activity metric," Patent EP2 070 048 B1, Nov., 2010, eP Patent 2,070,048.X. Lu, J. Zhai, and C. Gomila, "Spatial masking using a spatial activity metric," Patent EP2 070 048 B1, Nov., 2010, and EP Patent 2,070,048.

[0094] [34] K. Subbalakshmi and P. Amin, “Robust hidden data extraction method for scaling attacks," Patent US 20 060 176 299 A1, Aug., 2006, uS Patent App. 11/200,866.[0094] [34] K. Subbalakshmi and P. Amin, "Robust hidden data extraction method for scaling attacks," US Patent 20 060 176 299 A1, Aug., 2006, US Patent App. 11 / 200,866.

[0095] [35] H. Do Viet, D. C. T. Le, D. T. Nguyen, N. L. Nguyen, and T. T. H. Nguyen, “Apparatus and method for adaptive 3d artifact reducing for encoded image signal," Patent CA 2 616 875 A1, Feb., 2006, cA Patent App. CA 2,616,875.[0095] [35] H. Do Viet, DCT Le, DT Nguyen, NL Nguyen, and TTH Nguyen, "Apparatus and method for adaptive 3d artifact reducing for encoded image signal," Patent CA 2,616,875 A1, Feb., 2006 , cA Patent App. CA 2,616,875.

[0096] [36] B. Song, “Advanced DCT-based video encoding method and apparatus," Patent US 20 040 228 406 A1, Nov., 2004, uS Patent App. 10/642,646.[0096] [36] B. Song, "Advanced DCT-based video encoding method and apparatus," US Patent 20 040 228 406 A1, Nov., 2004, US Patent App. 10 / 642,646.

[0097] [37] D. C. T. Le, “Apparatus and method for adaptive spatial segmentation-based noise reducing for encoded image signal," Patent WO 2 002 102 086 A3, Feb., 2004, w O Patent App. PCT/CA2002/000.887.[0097] [37] DCT Le, "Apparatus and method for adaptive spatial segmentation-based noise reducing for encoded image signal," Patent WO 2 002 102 086 A3, Feb., 2004, The Patent App. PCT / CA2002 / 000.887 .

[0098] [38] T. Pohjola, “Encoding method and arrangement," Patent EP 1 324 618 A2, Jul., 2003, eP Patent App. EP20,020,396,159.[0098] [38] T. Pohjola, "Encoding Method and Arrangement," Patent EP 1,324,618 A2, Jul., 2003, and Patent App. EP20,020,396,159.

[0099] [39] T. Ono, “Technique of embedding and detecting digital watermark,” Patent US 7 123 741 B2, Jan., 2003, uS Patent App. 10/139,602.[0099] [39] T. Ono, "Technique of Embedding and Detecting Digital Watermark," Patent US 7 123 741 B2, Jan., 2003, US Patent App. 10 / 139,602.

[00100] [40] I. J. Cox, M. L. Miller, and R. Oami, “Robust digital watermarking," Patent CA 2 276 378 C, Feb., 2003, cA Patent 2,276,378.[00100] [40] I.J. Cox, M.L. Miller, and R. Oami, "Robust digital watermarking," Patent CA 2,276,378 C, Feb., 2003, cA Patent 2,276,378.

[00101] [41] I. Hajjahmad, M. L. Reisch, L. E. Sunshine, Μ. A. Wober. and Y. Yang, “Image noise reduction system using a wiener variant filter in a pyramid image representation," Patent EP 0 826 195 B1, Dec., 1998, eP Patent 0,826,195.[00101] [41] I. Hajjahmad, M.L. Reisch, L.E. Sunshine, Μ. A. Wober. and Y. Yang, "Image noise reduction system using a wiener variant filter in a pyramid image representation," Patent EP 0 826 195 B1, Dec., 1998, and EP Patent 0,826,195.

[00102] [42] Μ. A. Wober and M. L. Reisch, “Method for scaling and filtering images using discrete cosine transforms,” Patent EP 0 731 957 B1, Oct., 1997, eP Patent 0,731.957.[42] 42. A. Wober and M.L. Reisch, "Method for scaling and filtering images using discrete cosine transforms," Patent EP 0 731 957 B1, Oct., 1997, and EP Patent 0.731.957.

[00103] [43] R. L. Graham, D. E. Knuth, and O. Patashnik, Concrete Mathematics.Reading, MA: Addison-Wesley Publishing Company, 1989.[00103] [43] R. L. Graham, D. E. Knuth, and O. Patashnik, Concrete Mathematics.Reading, MA: Addison-Wesley Publishing Company, 1989.

[00104] [44] G. Boudreaux-Bartels and T. W. Parks, “Discrete Fourier transform using summation by parts," in IEEE International Conference on Acoustics, Speech, and Signal Processing, vol. 12, Apr. 1987, pp. 1827-1830.[00104] [44] G. Boudreaux-Bartels and TW Parks, “Discrete Fourier transform using summation by parts,” in IEEE International Conference on Acoustics, Speech, and Signal Processing, vol. 12, Apr. 1987, pp. 1827- 1830

[00105] [45] R. J. Cintra, “Soma por partes para sinais peri ódicos e transformadas discretas," in XXX Simpósio Brasileiro de Telecomunica.co^es -SBrT’12, Bras ília, DF, Sep. 2012.[00105] [45] R. J. Cintra, “Sum of Pieces for Discrete Transformed Periodic Signals," in XXX Brazilian Telecommunication Symposium-es-SBrT'12, Brasilia, DF, Sep. 2012.

[00106] [46] A. V. Oppenheim, R. W. Schafer, Μ. T. Yoder, and W. T. Padgett, Discrete Time Signal Processing, 3rd ed., A. V. Oppenheim, Ed. Upper Saddle River, NJ: Prentice-Hall, Inc., Aug. 2009, vol. 1.[46] A. V. Oppenheim, R. W. Schafer, Μ. T. Yoder, and W. T. Padgett, Discrete Time Signal Processing, 3rd ed., A. V. Oppenheim, Ed. Upper Saddle River, NJ: Prentice-Hall, Inc., Aug. 2009, vol. 1.

[00107] [47] R. C. Gonzalez and R. E. Woods, Digital Image Processing, 2nd ed.Prentice-Hall, Inc., 2001.R. C. Gonzalez and R. E. Woods, Digital Image Processing, 2nd ed.Prentice-Hall, Inc., 2001.

[00108] [48] R. W. Hamming, Digital Filters, 3rd ed. Hertfordshire, UK: Prentice Hall International Ltd., 1989.R. W. Hamming, Digital Filters, 3rd ed. Hertfordshire, UK: Prentice Hall International Ltd., 1989.

[00109] [49] H. El-Banna, A. A. El-Fattah, and W. Fakhr, “An efficient implementation of the 1D-DCT using FPGA technology," in Engineering of Computer-Based Systems, 2004. Proceedings. 11th IEEE International Conference and Workshop on the, 2004, pp. 356-360.[00109] [49] H. El-Banna, AA El-Fattah, and W. Fakhr, “An efficient implementation of 1D-DCT using FPGA technology,” in Engineering of Computer-Based Systems, 2004. Proceedings. 11th IEEE International Conference and Workshop on the, 2004, pp. 356-360.

[00110] [50] S. J. Shan, C. Chen, and E. Yang, “High performance 2-D IDCT for image/video decoding based on FPGA," in 2012 International Conference on Audio, Language and Image Processing (ICALIP), 2012, pp. 33-38.[00110] [50] SJ Shan, C. Chen, and E. Yang, “High-performance 2-D IDCT for image / video decoding based on FPGA," in 2012 International Conference on Audio, Language and Image Processing (ICALIP), 2012, pp. 33-38.

[00111] [51] H. K. C.-K. Fong and W.-K. Cham, “LLM integer cosine transform and its fast algorithm,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 22, no. 6,pp. 844-854, 2012.[51] H. K. C.-K. Fong and W.-K. Cham, “LLM Integral Cosine Transform and Its Fast Algorithm,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 22, no. 6, pp. 844-854, 2012.

[00112] [52] M. Abramowitz and I. A. Stegun, Eds., Handbook of Mathematical Functions: with Formulas, Graphs, and Mathematical Tables. MIT Press, 1964.[00112] [52] M. Abramowitz and I. A. Stegun, Eds., Handbook of Mathematical Functions: with Formulas, Graphs, and Mathematical Tables. MIT Press, 1964.

[00113] [53] E. Zheng, Z. Liu, and L. Ma, “Study on harmonic detection method based on FFT and wavelet transform," in 2nd International Conference on Signal Processing Systems (ICSPS), vol. 3, 2010.[00113] [53] E. Zheng, Z. Liu, and L. Ma, “Study on harmonic detection method based on FFT and wavelet transform,” in 2nd International Conference on Signal Processing Systems (ICSPS), vol. 3, 2010 .

[00114] [54] Mukherjee and S. K. Mitra, "Enhancement of color images by scaling the DCT coefficients,” IEEE Transactions on Image Processing, vol. 17, no. 10, pp. 1783-1794, 2008.[00114] [54] Mukherjee and S. K. Mitra, "Enhancement of color images by scaling the DCT coefficients," IEEE Transactions on Image Processing, vol. 17, no. 10, pp. 1783-1794, 2008.

[00115] [55] S. Bouguezel, M. O. Ahmad, and Μ. N. S. Swamy, “A multiplication-free transform for image compression," in 2nd International Conference on Signals, Circuits and Systems,Monastir, TN, 2008, pp. 1-4.[00115] [55] S. Bouguezel, M.O. Ahmad, and Μ. N. S. Swamy, “A multiplication-free transform for image compression,” in 2nd International Conference on Signals, Circuits and Systems, Monastir, TN, 2008, pp. 1-4.

[00116] [56] V. Dimitrov and K. A. Wahid, “On the error-free computation of fast cosine transform," Information Theories and Applications, vol. 12, no. 4, pp. 321-327, 2005.[00116] [56] V. Dimitrov and K. A. Wahid, “On the error-free computation of fast cosine transform," Information Theories and Applications, vol. 12, no. 4, pp. 321-327, 2005.

[00117] [57] P.. Duhamel and H. H'Mida, “New 2n DCT algorithms suitable for VLSI implementation.ln IEEE International Conference on Acoustics, Speech, and Signal Processing, vol. 12, 1987, pp. 1805-1808.[00117] [57] P. Duhamel and H. H'Mida, “New 2n DCT algorithms suitable for VLSI implementation. IEEE International Conference on Acoustics, Speech, and Signal Processing, vol. 12, 1987, pp. 1805-1808.

REIVINDICAÇÕES

Claims (22)

1. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula que faz o processamento de voz, áudio, imagens ou sinais que calcule os coeficientes da DCT de um sinal de média nula x—[n], apresentada de forma compacta por meio de representação matricial, caracterizado pelo uso do circuito esquemático “C” que compreende meios de fatoração de matriz de escalonamento “S”, com matriz de permutação “P”, com matriz simétrica “ΜΓ, com matriz ΈΓ, com matriz “R2”, com matriz “R3”, com matriz simétrica “M3", com matriz simétrica “M4", e com uma matriz aditiva “A", após que o bloco implemente a acumulação do sinal de entrada.1. A method for computing DCT spectrum for accumulated and / or null average signals that processes voice, audio, images, or signals that computes the DCT coefficients of a null average signal x— [n] shown in compact form by means of matrix representation, characterized by the use of the schematic circuit "C" which comprises scaling matrix factorization means "S", with permutation matrix "P", with symmetrical matrix "ΜΓ, with matrix ΈΓ, with matrix “R2”, with matrix “R3”, with symmetrical matrix “M3", with symmetrical matrix “M4", and with an additive matrix “A”, after the block implements the input signal accumulation. 2. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 1, caracterizado pela matriz de escalonamento “S” contribuir apenas com fatores de escalonamentos para a computação dos coeficientes da DCT.Method for computing the DCT spectrum for accumulated and / or zero mean signals according to claim 1, characterized in that the scaling matrix "S" contributes only scaling factors to compute the DCT coefficients. 3. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 1, caracterizado pelas matrizes “ΜΓ, “M3" e “M4" serem componentes de matriz diagonal “M".Method for computing the DCT spectrum for accumulated and / or zero mean signals according to claim 1, characterized in that the matrices "ΜΓ," M3 "and" M4 "are diagonal matrix components" M ". 4. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 1, caracterizado pela matriz “R1” conter multiplicandos que satisfazem a expressão: s2 + s6 igual a V2 que multiplica c2 e ainda s6 - s2 igual a V2 que multiplica s2.Method for computing the DCT spectrum for accumulated and / or zero mean signals according to claim 1, characterized in that the matrix "R1" contains multiplicands which satisfy the expression: s2 + s6 equal to V2 which multiplies c2 and further s6 - s2 equals V2 multiplying s2. 5. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 1, caracterizado pela matriz "R2" ser matriz diagonal e ser igual a diag(1, 1, s2, 1, 1, 1, 1,1).Method for computing the DCT spectrum for accumulated and / or null average signals according to claim 1, characterized in that the matrix "R2" is diagonal matrix and is equal to diag (1, 1, s2, 1, 1 , 1.1, 1.1). 6. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 1, caracterizado pela matriz “A” ser uma matriz aditiva que consiste que um estágio borboleta usual esteja presente em algoritmos de decimação em frequência.Method for computing the DCT spectrum for accumulated and / or null average signals according to claim 1, characterized in that the matrix "A" is an additive matrix consisting of a usual butterfly stage present in decimation algorithms in frequency. 7. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula que faz o processamento de voz, áudio, imagens ou sinais que calcule os coeficientes da DCT de um sinal de média nula x-[n], apresentada de forma compacta por meio de representação matricial, caracterizado pelo uso do circuito esquemático “C” que compreende meios de fatoração de matriz de escalonamento “S", com matriz de permutação “P", com matriz simétrica “ΜΓ, com matriz “Ri”, com matriz “R2”t com matriz “R3", com matriz simétrica “M3”, com matriz simétrica “M4", e com uma matriz aditiva “A”, após que 0 bloco implemente a acumulação do sinal de entrada;7. Method for computing DCT stepped spectrum for accumulated and / or null average signals that processes voice, audio, images, or signals that computes the DCT coefficients of a null average signal x- [n] shown compactly by means of matrix representation, characterized by the use of the schematic circuit "C" comprising scaling matrix factorization means "S", with permutation matrix "P", with symmetric matrix "ΜΓ, with matrix" Ri " , with matrix “R2” t with matrix “R3", with symmetrical matrix “M3”, with symmetrical matrix “M4", and with an additive matrix “A”, after 0 block implements the input signal accumulation; 8. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 7, caracterizado pelas matrizes “ΜΛ UM3" e “M4" serem componentes de matriz diagonal “M”.Method for computing the scaled DCT spectrum for accumulated and / or null average signals according to claim 7, characterized in that the matrices "ΜΛ UM3" and "M4" are diagonal matrix components "M". 9. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 7, caracterizado pela matriz “Rr conter multiplicandos que satisfazem a expressão: s2 + s6 igual a V2 que multiplica c2 e ainda s6 - s2 igual a V2 que multiplica s2.Method for computing the scaled DCT spectrum for accumulated and / or null average signals according to claim 7, characterized in that the matrix "Rr contains multiplands satisfying the expression: s2 + s6 equal to V2 multiplying c2 and further s6 - s2 equals V2 multiplying s2. 10. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 7, caracterizado pela matriz “R2” ser matriz diagonal e ser igual a diag(1, 1, s2, 1, 1.1. 1.1).Method for computing the scaled DCT spectrum for accumulated and / or null average signals according to claim 7, characterized in that the matrix "R2" is diagonal matrix and is equal to diag (1, 1, s2, 1, 1.1). 11. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 7, caracterizado pela matriz “A" ser uma matriz aditiva que consiste que um estágio borboleta usual esteja presente em algoritmos de decimação em frequência.Method for computing DCT stepped spectrum for accumulated and / or null average signals according to claim 7, characterized in that the matrix "A" is an additive matrix consisting of a usual butterfly stage present in decimation algorithms. in frequency. 12. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula que faz o processamento de voz, áudio, imagens ou sinais que calcule os coeficientes da DCT de um sinal x [n], apresentada de forma compacta por meio de representação matricial, caracterizado pelo uso do circuito esquemático “C” que compreende meios de fatoração de matriz de escalonamento “S”, com matriz de permutação “P”, com matriz simétrica “ΜΓ, com matriz “RT, com matriz “R2n, com matriz “R3”, com matriz simétrica “M3", com matriz simétrica “M4”, e com uma matriz aditiva “A", após bloco que implementa a remoção do valor DC do sinal x [n] do correspondente sinal acumulado z [n], resultando em z~[n].12. Method for computing the DCT spectrum for accumulated and / or null average signals which processes voice, audio, images or signals that computes the DCT coefficients of a signal x [n], presented compactly by means of matrix representation, characterized by the use of the schematic circuit “C” which comprises scaling matrix factorization means “S”, with permutation matrix “P”, with symmetric matrix “ΜΓ, with matrix” RT, with matrix “R2n, with matrix “R3”, with symmetrical matrix “M3", with symmetrical matrix “M4”, and with an additive matrix “A”, after block that implements the removal of the DC value of signal x [n] from the corresponding accumulated signal z [ n], resulting in z ~ [n]. 13. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 12, caracterizado pela matriz de escalonamento “S” contribuir apenas com fatores de escalonamentos para a computação dos coeficientes da DCT.Method for computing the DCT spectrum for accumulated and / or zero mean signals according to claim 12, characterized in that the scaling matrix "S" contributes only scaling factors to compute the DCT coefficients. 14. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 12, caracterizado pelas matrizes “M3” e “M4” serem componentes de matriz diagonal “M”.Method for computing the DCT spectrum for accumulated and / or null average signals according to claim 12, characterized in that the matrices "M3" and "M4" are diagonal matrix components "M". 15. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 12, caracterizado pela matriz ΈΓ conter multiplicandos que satisfazem a expressão: s2 + s6 igual a V2 que multiplica c2 e ainda s6 - s2 igual a V2 que multiplica s2.Method for computing the DCT spectrum for accumulated and / or null average signals according to claim 12, characterized in that the matrix ΈΓ contains multiplicands satisfying the expression: s2 + s6 equal to V2 which multiplies c2 and still s6 - s2 equals V2 which multiplies s2. 16. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 12, caracterizado pela matriz “R2” ser matriz diagonal e ser igual a diag(1, 1, s2. 1, 1,1,1,1).Method for computing the DCT spectrum for accumulated and / or null average signals according to claim 12, characterized in that the matrix "R2" is diagonal matrix and is equal to diag (1, 1, s2. 1, 1 , 1,1,1). 17. Método para computação do espectro da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 12, caracterizado pela matriz “A" ser uma matriz aditiva que consiste que um estágio borboleta usual esteja presente em algoritmos de decimação em frequência.Method for computing the DCT spectrum for accumulated and / or null average signals according to claim 12, characterized in that the matrix "A" is an additive matrix consisting of a usual butterfly stage present in decimation algorithms in frequency. 18. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula que faz o processamento de voz, áudio, imagens ou sinais que calcule os coeficientes da DCT de um sinal x [n], apresentada de forma compacta por meio de representação matricial, caracterizado pelo uso do circuito esquemático “C” que compreende meios de fatoração de matriz de escalonamento “S”, com matriz de per- mutação “P”, com matriz simétrica “ΜΓ, com matriz “Ri”, com matriz “R2", com matriz “R3”, com matriz simétrica “M3", com matriz simétrica “M4", e com uma matriz aditiva “A", após bloco que implementa a remoção do valor DC do sinal x [n] do correspondente sinal acumulado z [n], resultando em z_[n],18. A method for computing DCT stepped spectrum for accumulated and / or null average signals that processes voice, audio, images, or signals that computes the DCT coefficients of a signal x [n], presented compactly by matrix representation means, characterized by the use of the schematic circuit "C" comprising scaling matrix factorization means "S", with permutation matrix "P", symmetrical matrix "ΜΓ, matrix" Ri ", with matrix "R2", matrix "R3", symmetric matrix "M3", symmetric matrix "M4", and an additive matrix "A", after block that implements the removal of the DC value of signal x [n] from the corresponding accumulated signal z [n], resulting in z_ [n], 19. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 18, caracterizado pelas matrizes “ΜΓ, "M3” e “M4” serem componentes de matriz diagonal “M".A method for computing the DCT scaled spectrum for accumulated and / or null average signals according to claim 18, characterized in that the matrices "ΜΓ," M3 "and" M4 "are diagonal matrix components" M ". 20. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 18, caracterizado pela matriz “FV conter multiplicandos que satisfazem a expressão: s2 + s6 igual a V2 que multiplica c2 e ainda s6 - s2 igual a V2 que multiplica s2.Method for computing the scaled DCT spectrum for accumulated and / or null average signals according to claim 18, characterized in that the matrix "FV contains multiplicands satisfying the expression: s2 + s6 equal to V2 multiplying c2 and further s6 - s2 equals V2 multiplying s2. 21. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 18, caracterizado pela matriz “R2n ser matriz diagonal e ser igual a diag(1, 1, s2, 1, 1.1, 1,1).Method for computing the scaled DCT spectrum for accumulated and / or null average signals according to claim 18, characterized in that the matrix "R2n is diagonal matrix and is equal to diag (1, 1, s2, 1, 1.1"). , 1.1). 22. Método para computação do espectro escalonado da DCT para sinais acumulados e/ou de média nula, de acordo com a reivindicação 18, caracterizado pela matriz “A” ser uma matriz aditiva que consiste que um estágio borboleta usual esteja presente em algoritmos de decimação em frequência.Method for computing the scaled DCT spectrum for accumulated and / or null average signals according to claim 18, characterized in that the matrix "A" is an additive matrix consisting of a usual butterfly stage present in decimation algorithms. in frequency.
BR102015017870-0A 2015-07-27 2015-07-27 METHOD FOR COMPUTING THE DCT SPECTRUM FOR ACCUMULATED AND / OR MEDIUM NULL SIGNS BR102015017870A2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
BR102015017870-0A BR102015017870A2 (en) 2015-07-27 2015-07-27 METHOD FOR COMPUTING THE DCT SPECTRUM FOR ACCUMULATED AND / OR MEDIUM NULL SIGNS

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
BR102015017870-0A BR102015017870A2 (en) 2015-07-27 2015-07-27 METHOD FOR COMPUTING THE DCT SPECTRUM FOR ACCUMULATED AND / OR MEDIUM NULL SIGNS

Publications (1)

Publication Number Publication Date
BR102015017870A2 true BR102015017870A2 (en) 2017-11-07

Family

ID=62529564

Family Applications (1)

Application Number Title Priority Date Filing Date
BR102015017870-0A BR102015017870A2 (en) 2015-07-27 2015-07-27 METHOD FOR COMPUTING THE DCT SPECTRUM FOR ACCUMULATED AND / OR MEDIUM NULL SIGNS

Country Status (1)

Country Link
BR (1) BR102015017870A2 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113049080A (en) * 2021-03-08 2021-06-29 中国电子科技集团公司第三十六研究所 GDWC auditory feature extraction method for ship radiation noise

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113049080A (en) * 2021-03-08 2021-06-29 中国电子科技集团公司第三十六研究所 GDWC auditory feature extraction method for ship radiation noise

Similar Documents

Publication Publication Date Title
Porwik et al. The Haar-wavelet transform in digital image processing: its status and achievements
Wang et al. A novel synchronization invariant audio watermarking scheme based on DWT and DCT
Bradley Shift-invariance in the discrete wavelet transform
Raviraj et al. The modified 2D-Haar wavelet transformation in image compression
Song et al. 2D Gabor filters-based steganalysis of content-adaptive JPEG steganography
Mehta et al. A hybrid semi-blind gray scale image watermarking algorithm based on DWT-SVD using human visual system model
Coelho et al. Efficient computation of the 8-point DCT via summation by parts
BR102015017870A2 (en) METHOD FOR COMPUTING THE DCT SPECTRUM FOR ACCUMULATED AND / OR MEDIUM NULL SIGNS
JP2019061671A (en) Block-based principal component analysis transformation method and device thereof
Chen et al. Adaptive digital watermarking using RDWT and SVD
Dagher et al. Image denoising using fourth order wiener filter with wavelet quadtree decomposition
Kala et al. Robust lossless image watermarking in integer wavelet domain using SVD
KR100633555B1 (en) Signal processing
Bhat K et al. Audio watermarking based on quantization in wavelet domain
Shin Fast detection of copy-move forgery image using three step search algorithm in the spatial domain
Ananthaneni et al. Hybrid digital image watermarking using contourlet transform (CT), DCT and SVD
Shi et al. Detection of audio interpolation based on singular value decomposition
Ng et al. An improved image steganography scheme based on RDWT and QR decomposition
Hong et al. New fast DCT algorithms based on Loeffler's factorization
Nha et al. An improved QR decomposition for color image watermarking
Pandey Analysis of image compression using wavelets
Mukherjee et al. Image filtering in the compressed domain
Lien et al. A Novel Image Super-Resolution Technology Based on the Wavelet Coefficients Prediction Scheme
Ke et al. Hybrid Domain Audio Watermarking Embedding Algorithm Based on Redundant Discrete Wavelet Transform
Tagliasacchi et al. Transform coder identification

Legal Events

Date Code Title Description
B03A Publication of a patent application or of a certificate of addition of invention [chapter 3.1 patent gazette]
B11A Dismissal acc. art.33 of ipl - examination not requested within 36 months of filing
B04C Request for examination: application reinstated [chapter 4.3 patent gazette]
B06F Objections, documents and/or translations needed after an examination request according [chapter 6.6 patent gazette]
B08F Application dismissed because of non-payment of annual fees [chapter 8.6 patent gazette]

Free format text: REFERENTE A 5A ANUIDADE.

B08G Application fees: restoration [chapter 8.7 patent gazette]
B08F Application dismissed because of non-payment of annual fees [chapter 8.6 patent gazette]

Free format text: REFERENTE A 6A ANUIDADE.

B08G Application fees: restoration [chapter 8.7 patent gazette]
B08F Application dismissed because of non-payment of annual fees [chapter 8.6 patent gazette]

Free format text: REFERENTE A 7A ANUIDADE.

B08G Application fees: restoration [chapter 8.7 patent gazette]
B08F Application dismissed because of non-payment of annual fees [chapter 8.6 patent gazette]

Free format text: REFERENTE A 8A ANUIDADE.

B08G Application fees: restoration [chapter 8.7 patent gazette]
B06V Preliminary requirement: patent application procedure suspended [chapter 6.22 patent gazette]
B11B Dismissal acc. art. 36, par 1 of ipl - no reply within 90 days to fullfil the necessary requirements
B11N Dismissal: publication cancelled [chapter 11.14 patent gazette]

Free format text: ANULADA A PUBLICACAO CODIGO 11.2 NA RPI NO 2761 DE 05/12/2023 POR TER SIDO INDEVIDA.

B09B Patent application refused [chapter 9.2 patent gazette]