Patents by Inventor Toshihiro Gai
Toshihiro Gai has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Patent number: 8218083Abstract: A video noise reducer processes a frame difference signal equal to the difference between the video signal of the current frame and the video signal of the preceding frame or the difference between the video signal of the current frame and the noise-reduced video signal of the preceding frame to obtain a motion detection signal and a noise detection signal. The difference between the motion detection signal and the noise detection signal is then processed to obtain a recursion coefficient. The frame difference signal is multiplied by the recursion coefficient and the resulting product is additively combined with the video signal of the current frame to reduce noise without generating significant motion artifacts.Type: GrantFiled: March 16, 2007Date of Patent: July 10, 2012Assignee: Mitsubishi Electric CorporationInventors: Masako Asamura, Koji Minami, Chihiro Sakuwa, Toshihiro Gai
-
Patent number: 8218632Abstract: A frame interpolation apparatus generates a pixel value at an interpolation position by examining pairs of pixel blocks, one from a preceding reference frame and one from a following reference frame, located in positions that are point-symmetric or nearly point-symmetric with respect to the interpolation position. The pair of pixel blocks showing the greatest mutual similarity is selected, the central pixels in the selected pair of pixel blocks are taken as reference pixels, and the interpolated pixel value is generated from the reference pixel values. By routinely examining a large number of pairs of pixel blocks for each interpolation position, the frame interpolation apparatus can detect image motion accurately and generate accurate interpolated values.Type: GrantFiled: September 19, 2007Date of Patent: July 10, 2012Assignee: Mitsubishi Electric CorporationInventors: Koji Minami, Toshihiro Gai, Hironobu Yasui, Shuji Sotoda, Chihiro Sakuwa, Takashi Fujiwara
-
Patent number: 8174614Abstract: A video signal processing apparatus, which can prevent degradation of picture quality in converting an interlaced scan video signal to a progressive scan video signal, includes an interframe correlation detection means (11); an interfield correlation detection means (12); a field resolution determination means (13) which generates a signal (filhv) indicative of a degree of high frequency components existing in a field; a pull-down sequence detection means (14) which generates a telecine detection signal (tci) indicating whether or not an input video signal is any of 2-3 pull-down telecine video signal and 2-2 pull-down telecine video signal and also generates a mixing ratio signal (tcmix) indicative of a mixing ratio used at the time of interpolation processing; an interpolation signal generation means (110) which mixes a plurality of kinds of interpolation signals (Im, It) in accordance with the telecine detection signal (tci) and the mixing ratio signal (tcmix); and a rate doubling conversion means (9) whicType: GrantFiled: November 21, 2005Date of Patent: May 8, 2012Assignee: Mitsubishi Electric CorporationInventors: Masako Asamura, Koji Minami, Chihiro Sakuwa, Toshihiro Gai
-
Patent number: 8150204Abstract: A noise reducer for a signal determines a direction of correlation for each picture element (pixel) by calculating sums of absolute differences between a block centered on the pixel and neighboring blocks centered on surrounding pixels and choosing the direction of the neighboring block with the least sum of absolute differences. The pixel is then filtered one-dimensionally by averaging it with neighboring pixels located in the direction of correlation. This noise reduction method is simple, reliable, and avoids needless loss of detail.Type: GrantFiled: March 7, 2008Date of Patent: April 3, 2012Assignee: Mitsubishi Electric CorporationInventors: Takashi Fujiwara, Toshihiro Gai, Koji Minami, Hironobu Yasui
-
Patent number: 7667721Abstract: A gray-scale conversion means (10) converts the gray scale of an input video signal (100) as necessary, and outputs the video signal with the converted gray scale. A statistical processing means (13) detects a statistical property of each field or frame of the input video signal; a scene change detection means (16) detects scene changes in the input video signal. A conversion characteristic determination means (22) determines the gray-scale conversion characteristic according to outputs of the statistical processing means (13) and the scene change detection means (16). The gray-scale correction increases contrast regardless of image content, and enables faster response to changes in image content.Type: GrantFiled: February 16, 2005Date of Patent: February 23, 2010Assignee: Mitsubishi Electric CorporationInventors: Koji Minami, Chihiro Sakuwa, Toshihiro Gai, Masako Asamura
-
Patent number: 7515211Abstract: A video signal processing circuit that uses a prescribed clock signal to process a digitized composite video signal. A clock generating means (2) generates the prescribed clock signal; a burst phase detecting means (3) detects color subcarrier phase information (p) in each line of the composite video signal; a phase difference calculation means (4) finds the phase difference between phase information (p) from the burst phase detecting means and a prescribed reference phase; a sampling phase conversion means (8) corrects the sampling phase of the composite video signal according to phase corrections (?b, ?t) obtained from the phase difference calculation means (4); a Y/C separation means (9) separates the luminance and chrominance signals from the composite video signal output from the sampling phase conversion means (8). Excellent two- or three-dimensional Y/C separation can be obtained regardless of the television broadcast system, even from a non-standard signal.Type: GrantFiled: August 25, 2004Date of Patent: April 7, 2009Assignee: Mitsubishi Denki Kabushiki KaishaInventors: Masako Asamura, Yoshito Suzuki, Toshihiro Gai, Koji Minami, Masaki Yamakawa
-
Publication number: 20080231753Abstract: A noise reducer for a signal determines a direction of correlation for each picture element (pixel) by calculating sums of absolute differences between a block centered on the pixel and neighboring blocks centered on surrounding pixels and choosing the direction of the neighboring block with the least sum of absolute differences. The pixel is then filtered one-dimensionally by averaging it with neighboring pixels located in the direction of correlation. This noise reduction method is simple, reliable, and avoids needless loss of detail.Type: ApplicationFiled: March 7, 2008Publication date: September 25, 2008Inventors: Takashi Fujiwara, Toshihiro Gai, Koji Minami, Hironobu Yasui
-
Publication number: 20080151103Abstract: A video signal processing apparatus, which can prevent degradation of picture quality in converting an interlaced scan video signal to a progressive scan video signal, includes an interframe correlation detection means (11); an interfield correlation detection means (12); a field resolution determination means (13) which generates a signal (filhv) indicative of a degree of high frequency components existing in a field; a pull-down sequence detection means (14) which generates a telecine detection signal (tci) indicating whether or not an input video signal is any of 2-3 pull-down telecine video signal and 2-2 pull-down telecine video signal and also generates a mixing ratio signal (tcmix) indicative of a mixing ratio used at the time of interpolation processing; an interpolation signal generation means (110) which mixes a plurality of kinds of interpolation signals (Im, It) in accordance with the telecine detection signal (tci) and the mixing ratio signal (tcmix); and a rate doubling conversion means (9) whicType: ApplicationFiled: November 21, 2005Publication date: June 26, 2008Inventors: Masako Asamura, Koji Minami, Chihiro Sakuwa, Toshihiro Gai
-
Publication number: 20080069217Abstract: A frame interpolation apparatus generates a pixel value at an interpolation position by examining pairs of pixel blocks, one from a preceding reference frame and one from a following reference frame, located in positions that are point-symmetric or nearly point-symmetric with respect to the interpolation position. The pair of pixel blocks showing the greatest mutual similarity is selected, the central pixels in the selected pair of pixel blocks are taken as reference pixels, and the interpolated pixel value is generated from the reference pixel values. By routinely examining a large number of pairs of pixel blocks for each interpolation position, the frame interpolation apparatus can detect image motion accurately and generate accurate interpolated values.Type: ApplicationFiled: September 19, 2007Publication date: March 20, 2008Inventors: Koji Minami, Toshihiro Gai, Hironobu Yasui, Shuji Sotoda, Chihiro Sakuwa, Takashi Fujiwara
-
Publication number: 20080062101Abstract: A gray-scale conversion means (10) converts the gray scale of an input video signal (100) as necessary, and outputs the video signal with the converted gray scale. A statistical processing means (13) detects a statistical property of each field or frame of the input video signal; a scene change detection means (16) detects scene changes in the input video signal. A conversion characteristic determination means (22) determines the gray-scale conversion characteristic according to outputs of the statistical processing means (13) and the scene change detection means (16). The gray-scale correction increases contrast regardless of image content, and enables faster response to changes in image content.Type: ApplicationFiled: February 16, 2005Publication date: March 13, 2008Inventors: Koji Minami, Chihiro Sakuwa, Toshihiro Gai, Masako Asamura
-
Patent number: 7280709Abstract: In the conversion of a video signal from interlaced to progressive scanning, the value of each pixel on an interpolated scan line is calculated by a procedure that includes calculating similarity values for pairs of pixel blocks located in point-symmetrical positions on opposite sides of the interpolated pixel; deciding whether similar edges are present in corresponding positions in the two pixel blocks constituting each pixel block pair; selecting an interpolation direction corresponding to the most similar pixel block pair among the pixel block pairs in which the similar edges are present; and using the pixels disposed at or closest to the centers of the two pixel blocks in this pixel block pair as reference pixels. Restricting diagonal interpolation directions to pixel block pairs in which similar edges are present improves the accuracy of the interpolation direction.Type: GrantFiled: December 23, 2003Date of Patent: October 9, 2007Assignee: Mitsubishi Electric CorporationInventors: Koji Minami, Masako Asamura, Chihiro Sakuwa, Toshihiro Gai
-
Publication number: 20070229709Abstract: A video noise reducer processes a frame difference signal equal to the difference between the video signal of the current frame and the video signal of the preceding frame or the difference between the video signal of the current frame and the noise-reduced video signal of the preceding frame to obtain a motion detection signal and a noise detection signal. The difference between the motion detection signal and the noise detection signal is then processed to obtain a recursion coefficient. The frame difference signal is multiplied by the recursion coefficient and the resulting product is additively combined with the video signal of the current frame to reduce noise without generating significant motion artifacts.Type: ApplicationFiled: March 16, 2007Publication date: October 4, 2007Inventors: Masako Asamura, Koji Minami, Chihiro Sakuwa, Toshihiro Gai
-
Publication number: 20060290818Abstract: A video signal processing circuit that uses a prescribed clock signal to process a digitized composite video signal. A clock generating means (2) generates the prescribed clock signal; a burst phase detecting means (3) detects color subcarrier phase information (p) in each line of the composite video signal; a phase difference calculation means (4) finds the phase difference between phase information (p) from the burst phase detecting means and a prescribed reference phase; a sampling phase conversion means (8) corrects the sampling phase of the composite video signal according to phase corrections (?b, ?t) obtained from the phase difference calculation means (4); a Y/C separation means (9) separates the luminance and chrominance signals from the composite video signal output from the sampling phase conversion means (8). Excellent two- or three-dimensional Y/C separation can be obtained regardless of the television broadcast system, even from a non-standard signal.Type: ApplicationFiled: August 25, 2004Publication date: December 28, 2006Inventors: Masako Asamura, Yoshito Suzuki, Toshihiro Gai, Koji Minami, Masaki Yamakawa
-
Patent number: 7027098Abstract: A digitized luminance signal, representing the luminance component of a picture signal, is adjusted according to data stored in a memory. The data are output according to the value of the digitized luminance signal and multiplied by a weighting signal generated according to the amount of black area represented by the digitized luminance signal. The digitized luminance signal is then multiplied by a value obtained from the weighted product data. Use of the weighting signal enables the luminance scale to be stretched and compressed according to various different control characteristics without the need to store separate data for each characteristic in the memory.Type: GrantFiled: February 27, 2002Date of Patent: April 11, 2006Assignee: Mitsubishi Denki Kabushiki KaishaInventors: Toshihiro Gai, Junji Suzuki
-
Patent number: 6937287Abstract: In order to reduce the total capacity of a ROM used for adjustment of the black level and the white level of a digitized luminance signal for a television receiver, for example, a comparator compares the input luminance signal with a predetermined threshold value, a NOT gate inverts the input luminance signal, a first switching unit selects the input luminance signal or the output of the NOT gate, a ROM outputs an adjustment characteristic signal responsive to the output of the first switching unit, a multiplier multiplies the output of the ROM by a weighting signal and outputs the result of the multiplication as an adjustment value, a subtractor subtracts the adjustment value from the input luminance signal, an adder adds the adjustment value to the input luminance signal, and a second switching unit selects the output of the subtractor or the output of the adder.Type: GrantFiled: March 26, 2002Date of Patent: August 30, 2005Assignee: Mitsubishi Denki Kabushiki KaishaInventors: Toshihiro Gai, Yoshiaki Okuno, Jun Someya, Masaki Yamakawa
-
Publication number: 20040246546Abstract: In the conversion of a video signal from interlaced to progressive scanning, the value of each pixel on an interpolated scan line is calculated by a procedure that includes calculating similarity values for pairs of pixel blocks located in point-symmetrical positions on opposite sides of the interpolated pixel; deciding whether similar edges are present in corresponding positions in the two pixel blocks constituting each pixel block pair; selecting an interpolation direction corresponding to the most similar pixel block pair among the pixel block pairs in which the similar edges are present; and using the pixels disposed at or closest to the centers of the two pixel blocks in this pixel block pair as reference pixels. Restricting diagonal interpolation directions to pixel block pairs in which similar edges are present improves the accuracy of the interpolation direction.Type: ApplicationFiled: December 23, 2003Publication date: December 9, 2004Inventors: Koji Minami, Masako Asamura, Chihiro Sakuwa, Toshihiro Gai
-
Publication number: 20030086020Abstract: In order to reduce the total capacity of a ROM used for adjustment of the black level and the white level of a digitized luminance signal for a television receiver, for example, a comparator compares the input luminance signal with a predetermined threshold value, a NOT gate inverts the input luminance signal, a first switching unit selects the input luminance signal or the output of the NOT gate, a ROM outputs an adjustment characteristic signal responsive to the output of the first switching unit, a multiplier multiplies the output of the ROM by a weighting signal and outputs the result of the multiplication as an adjustment value, a subtractor subtracts the adjustment value from the input luminance signal, an adder adds the adjustment value to the input luminance signal, and a second switching unit selects the output of the subtractor or the output of the adder.Type: ApplicationFiled: March 26, 2002Publication date: May 8, 2003Inventors: Toshihiro Gai, Yoshiaki Okuno, Jun Someya, Masaki Yamakawa
-
Publication number: 20020196371Abstract: A digitized luminance signal, representing the luminance component of a picture signal, is adjusted according to data stored in a memory. The data are output according to the value of the digitized luminance signal and multiplied by a weighting signal generated according to the amount of black area represented by the digitized luminance signal. The digitized luminance signal is then multiplied by a value obtained from the weighted product data. Use of the weighting signal enables the luminance scale to be stretched and compressed according to various different control characteristics without the need to store separate data for each characteristic in the memory.Type: ApplicationFiled: February 27, 2002Publication date: December 26, 2002Inventors: Toshihiro Gai, Junji Suzuki
-
Patent number: 5517255Abstract: In a YC separating filter, a first horizontal chrominance extraction filter extracts frequency components corresponding to components of a color subcarrier in a horizontal direction to produce a first chrominance signal, a vertical chrominance extraction filter extracts frequency components corresponding to components of a color subcarrier in a vertical direction to produce a second chrominance signal, a horizontal and vertical chrominance extraction filter extracts frequency components corresponding to components of a color subcarrier in horizontal and vertical directions to produce a third chrominance signal, a first selector selects one of the first, second and third chrominance signals, a second horizontal chrominance extraction filter having a narrower bandwidth than the first horizontal chrominance extraction filter is provided to receive the selected chrominance signal, and a second selector selects the output of the second horizontal chrominance extraction filter when the first selector selects eitherType: GrantFiled: April 25, 1994Date of Patent: May 14, 1996Assignee: Mitsubishi Denki Kabushiki KaishaInventors: Toshihiro Gai, Tomonori Ohashi, Miyoichi Watanabe
-
Patent number: 5416531Abstract: A brightness signal/color signal separating filter. Non-correlative energies are extracted in various combinations of the horizontal scanning direction, the vertical scanning direction, a brightness signal and a color signal so as to judge the degree of video correlation in the horizontal direction and in the vertical direction.Type: GrantFiled: April 15, 1994Date of Patent: May 16, 1995Assignee: Mitsubishi Denki Kabushiki KaishaInventor: Toshihiro Gai