Megachips Corporation

Megachips Lsi Solutions Inc.

Yusuke Mizuno

Yujiro Tani

Hiromu Hasegawa

Masahiro Moriyama

Hideki Daian

Nobuhiro Minami

Gen Sasaki

Yuki Haraguchi

Takashi Mori

Shohei Nomoto

Yusuke Nara

Katsuya Hashimoto

Daisuke Togo

An ROI in a first image and a second image are synthesized by synthesizing first wavelet coefficient data corresponding to the first image and second wavelet coefficient data corresponding to the second image. An ROI coefficient and a non-ROI coefficient in the first wavelet coefficient data are determined on the basis of a mask for the first wavelet coefficient data, and the ROI coefficient and a wavelet coefficient in the second wavelet coefficient data, which is corresponding to the non-ROI coefficient, are synthesized. When the above coefficient synthesis is performed after a decomposition level of the first wavelet coefficient data is increased, the first wavelet coefficient data is adjusted on the basis of auxiliary wavelet coefficient data corresponding to an auxiliary image which can be equated with the first image before increasing the decomposition level.

Image processing system and image processing method

Wavelet transformation is performed on first image data and second image data until a decomposition level becomes a decomposition level based on synthesis control data or the like, and first wavelet coefficient data and second wavelet coefficient data are thereby generated. An ROI coefficient related to an ROI and a non-ROI coefficient in the first wavelet coefficient data are determined on the basis of mask data and the ROI coefficient in the first wavelet coefficient data and a wavelet coefficient in the second wavelet coefficient data are synthesized with each other, and synthesized coefficient data are thereby generated. Inverse wavelet transformation is performed on the synthesized coefficient data until a decomposition level becomes a predetermined end level, and synthetic image data are thereby generated.

An ROI coefficient and a non-ROI coefficient in first wavelet coefficient data corresponding to a first target image are determined on the basis of mask data which is developed for the first wavelet coefficient data. The ROI coefficient in the first wavelet coefficient data and a coefficient in second wavelet coefficient data corresponding to a second target image are synthesized. Synthesized coefficient data are thereby generated. Inverse wavelet transformation is performed on the synthesized coefficient data until a decomposition level becomes a predetermined end level. Synthetic image data are thereby generated.

Circuitry configured to generate an ROI mask by determining whether each of quantized wavelet coefficients constituting quantized image data is associated with an ROI or a non-ROI in an original image on the basis of the scaling amount of the Max-shift method, generate first image data by performing inverse quantization on the quantized image data after being scaled down, generate second image data having a specified decomposition level by performing inverse wavelet transformation on the first image data, generate a restored ROI mask having the same decomposition level as that of the second image data by performing a predetermined process on the ROI mask once or a plurality of times, generate masked image data by applying the restored ROI mask to the second image data, and perform the inverse wavelet transformation on the masked image data until a decomposition level becomes zero.

Image decoding apparatus and image decoding method

A very long instruction word (VLIW) processor that performs efficient processing including extended bits operations is provided. The VLIW processor includes an instruction control unit, a register file unit, and an instruction execution unit. The instruction execution unit includes a plurality of slots, and a state register arranged between the second slot and the third slot to transfer N-bit data between the second and third slots. The VLIW processor stores data output from the third slot into the state register and uses the data, and thus achieves efficient processing including bit-expanded operations, such as processing performed in response to instructions commonly used in image processing, image recognition, and other processing, while preventing scaling up of the circuit.

VLIW processor including a state register for inter-slot data transfer and extended bits operations

A WT unit performs wavelet transformation on original image data to generate first image data. An ROI developing unit, based on ROI information, specifies an ROI corresponding portion corresponding to an ROI and a non-ROI corresponding portion corresponding to a non-ROI to the first image data. A high-frequency cutting unit performs, on the first image data, a high-frequency cutting process that cuts a high-frequency component of the non-ROI corresponding portion. A low-frequency blurring unit performs, on the first image data, a low-frequency blurring process that blurs a low-frequency component of the non-ROI corresponding portion. The IWT unit performs inverse wavelet transformation on second image data (first image data obtained after the high-frequency cutting process and the low-frequency blurring process are performed) to generate third image data.

Image processing apparatus and image processing method

An image recognition apparatus determines whether an image of a pedestrian is captured in a frame of video data captured by a vehicle mounted camera. A pre-processing unit determines a detection block from within a frame, and cuts out block image data corresponding to the detection block from the frame. Block data with a predetermined size that is smaller than the size of the detection block is created from the block image data. A neuro calculation unit executes neuro calculation on the block data, and calculates an output synapse. A post-processing unit determines whether a pedestrian exists within the detection block on the basis of the output synapse. When a pedestrian is detected, the post-processing unit creates result data, which is obtained by superimposing the detection block within which the pedestrian was detected onto the frame.

Image recognition apparatus using neural network processing

An image scaling processor includes: a coefficient computing circuit that calculates interpolation coefficients to be used in an image scaling process; a multiplier that multiplies input image data by the interpolation coefficients provided from the coefficient computing circuit such that the interpolation coefficients respectively correspond to input pixels constituting the input image data; an adder that iteratively adds pieces of multiplied data output from the multiplier and obtains a total sum of the pieces of multiplied data for a predetermined number of the input pixels; a selector that outputs a total sum of the multiplied data at a timing at which the total sum of the pieces of multiplied data is obtained for the predetermined number of the input pixels; and a shift circuit that shifts an output of the selector to adjust a bit count of the output image data to a bit count of the input image data.

Image scaling processor and image scaling processing method

In a human detection device, an edge extractorcarries out edge extraction processing to an input imageand produces a horizontal edge image. A shoulder detectordetects a shoulder center and a shoulder width of a person included in the input image. A foot detectordetects a foot position of the person based on the detected shoulder center and shoulder width. A top detectordetects a top position of the person based on the detected shoulder center and shoulder width. A size determinerdetermines a horizontal size of the person based on the detected shoulder width and determines a vertical size of the person based on the detected foot position and top position. The size determinerproduces human range dataincluding the determined sizes, the shoulder center position, the foot position, and the top position.

Human detection device

An image scaling processor includes a coefficient table storing interpolation coefficients, a multiplier that multiplies each of pieces of pixel data constituting input image data and each of the interpolation coefficients stored in the coefficient table, an adder that repeatedly adds pieces of multiplied data output from the multiplier and obtains a total sum of the pieces of multiplied data for a predetermined number of the pieces of pixel data, a selector that outputs a total sum of the pieces of multiplied data at a timing at which the total sum of the pieces of multiplied data is obtained for the predetermined number of the pieces of pixel data, and a shift circuit that shifts an output of the selector to adjust a bit count of the output image data to a bit count of the input image data.

Method for creating coefficient table and image scaling processor

An image processing apparatus includes a first storage section and a second storage section, a storage control section, and a computation section. The storage control section sequentially acquires block images obtained as a result of dividing an input image, and stores the block image as a target block image in the first storage section, while storing, in the second storage section, image data of, in a region of the target block image, a region abutting un-inputted block images in the input image, as image data of an abuttal region. The computation section implements a resizing process for changing the size of the target block image by performing an interpolation calculation using image data of the target block image stored in the first storage section and the image data of the abuttal region stored in the second storage section.

Image resizing apparatus and method that interpolates image blocks with abuttal regions

A rate control unit performs data truncation on a code sequence having been sorted and bit shifted as shown in FIG.so that a desired noise reduction effect is achieved. The data truncation occurs in sequence from the rightmost bit. For example, in FIG., data is truncated from data of bitin a subband VHLdownwardly in sequence through data of bitin a subband YHH, and so on. If the desired noise reduction effect can be achieved by truncation of up to bit data in the subband YHH, data in those subbands in a dotted area in FIG.will be truncated. If the desired noise reduction effect cannot be achieved by truncation of up to bit data in the subband YHH, data will then be truncated from data of bitin a subband VLLdownwardly in sequence.

Compression encoder, compression encoding method and program

The image processorincludes a frequency transform unit, an encoding unitand a memory. The encoding unitincludes a DC processing unitthat generates a direct-current stream, an LP processing unitthat generates a low-frequency stream, an HP processing unitthat generates an upper high-frequency stream and a lower high-frequency stream, and an output unithaving output portstoto output the direct-current stream, the low-frequency stream, the upper high-frequency stream, and the lower high-frequency stream to the memory

Image processor

A sorting unit sorts a plurality of data sets of HP component having been processed by a decoding unit, selectively employing one of a first table corresponding to a first orientation of prediction and a second table corresponding to a second orientation of prediction in accordance with an orientation of prediction of HP component. The sorting unit includes an inverse prediction unit performing inverse prediction on data of LP component inputted from the decoding unit, a processing unit obtaining an orientation of prediction of HP component, based on the data of LP component after inverse prediction by the inverse prediction unit, and a selecting unit selecting one of the first and second tables, based on the orientation of prediction of HP component obtained by the processing unit.

Image decoder for HD photo

An image processor includes an encoding unit encoding inputted data. The encoding unit includes a first processing unit splitting the data into a first partial data in a first digit range on an upper side and a second partial data in a second digit range on a lower side, a second processing unit encoding only the first partial data between the first partial data and the second partial data, and a third processing unit performing correction to set a value of the first partial data at '0'.

Growing community of inventors

Osaka, Japan

Yusuke Mizuno