JP2011199712A - Video processor and video display device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that the gradation of corrected video is not the same as the gradation of uncorrected video if a filter coefficient is operated to process video.SOLUTION: This invention relates to a video processor including: a deformation ratio deriving part for deriving the deformation ratio of deformed video to original video about the deformed video in the case of executing deformation processing of the original image; a filter coefficient storage part for storing a plurality of filter coefficients in association with respective predetermined discrete deformation ratios; a filter coefficient selection part for selecting a filter coefficient corresponding to a plurality of deformation ratios close to the derived deformation ratio; a filter coefficient interpolating part for generating an interpolation filter coefficient derived by being interpolated on the basis of a filter coefficient selected in association with the plurality of deformation ratios; a filter coefficient adjusting part for adjusting the interpolation filter coefficient to generate an adjusted filter coefficient; and a pixel interpolating part for deriving and outputting the pixel value of each pixel of the deformed video by pixel interpolation by using the adjusted filter coefficient.

Description

  The present invention relates to a video processing apparatus.

  When displaying a rectangular image (hereinafter also referred to as an original image) on a projection surface such as a screen using a projector, the image displayed on the projection surface (hereinafter referred to as a projection image) depending on the relative positional relationship between the projector and the projection surface. (Also referred to as an image) may be distorted into a trapezoid, a parallelogram, or other quadrangle. Including the projection distortion other than the trapezoidal distortion, it is referred to as “trapezoidal distortion” in this specification. In this way, when trapezoidal distortion occurs in the projected image, a technique for correcting trapezoidal distortion that corrects the projected image to be displayed in a rectangular shape by using a projective transformation technique is used.

  In a projector that generates image light representing an image using a liquid crystal panel, an image distorted in the opposite direction to the projected image on the projection surface (hereinafter also referred to as a corrected image) when correcting keystone distortion, Produced on a liquid crystal panel. That is, the image formed on the liquid crystal panel is a rectangular image distorted into a trapezoid, a parallelogram, or other quadrangle, etc. The deformation rate is partially different. Therefore, a technique for calculating and setting a filter coefficient according to the deformation rate has been proposed (see, for example, Patent Document 1).

JP 11-341276 A

  However, when the image is processed by calculating the filter coefficient in this way, a problem has been found that the gradation (density) of the image after correction is not the same as the gradation of the image before correction.

  Note that such a problem is not limited to a projector, and is a problem common to video processing apparatuses capable of performing processing for transforming video.

  SUMMARY An advantage of some aspects of the invention is to solve at least a part of the problems described above, and the invention can be implemented as the following forms or application examples.

Application Example 1 A video processing apparatus,
A deformation rate deriving unit for deriving a deformation rate of the post-deformation video with respect to the original video for pixels in the post-deformation video when the original video is subjected to deformation processing based on a predetermined rule;
A storage unit for storing a filter coefficient for deriving a pixel value of the post-deformation video by pixel interpolation using a pixel value of the original video, and a plurality of the filters for each predetermined discrete deformation rate A filter coefficient storage unit that associates and stores coefficients;
A filter coefficient selection unit that selects the plurality of filter coefficients based on the deformation rate for each pixel of the post-deformation image, wherein the derived deformation rate is stored in the filter coefficient storage unit. A filter coefficient selecting unit that selects filter coefficients corresponding to a plurality of deformation ratios close to the derived deformation ratio,
In the filter coefficient selection unit, when filter coefficients corresponding to a plurality of deformation ratios are selected, the derived deformation ratio is obtained by performing interpolation based on the filter coefficients selected corresponding to the plurality of deformation ratios. A filter coefficient interpolation unit that generates a filter coefficient corresponding to
An interpolation result determination unit that determines whether or not an interpolation filter coefficient that is a filter coefficient derived in the filter coefficient interpolation unit satisfies a predetermined condition;
In the interpolation result determination unit, when it is determined that the predetermined condition is not satisfied, the interpolation filter coefficient is adjusted and a filter coefficient adjustment unit that generates an adjustment filter coefficient and the adjustment filter coefficient A pixel interpolation unit that derives and outputs a pixel value of each pixel of the transformed video by interpolation;
A video processing apparatus.

  According to this configuration, for example, even when the sum of the filter coefficients does not become 1 due to an error due to a fixed-point operation, pixel interpolation is performed using the adjustment filter coefficient, so that a preferable modified image is obtained. By adjusting the filter coefficient, a preferable post-deformation image can be obtained.

Application Example 2 The image processing apparatus according to Application Example 1, wherein the deformation process is a trapezoidal distortion correction process.
In this way, a preferable image can be obtained as the image after the trapezoidal distortion correction process.

Application Example 3 In the video processing device according to Application Example 1 or 2,
The predetermined condition is:
A video processing apparatus in which a sum of a plurality of the interpolation filter coefficients corresponding to the derived deformation rate is 1.

  By generating the adjustment filter coefficient so that the total sum of the filter coefficients becomes 1, pixel interpolation can be performed using an appropriate filter coefficient, so that deterioration in image quality can be suppressed.

Application Example 4 In the video processing device according to any one of Application Examples 1 to 3,
The image processing apparatus, wherein the deformation rate deriving unit derives the deformation rate for each pixel of the image after deformation.
According to this configuration, a more appropriate adjustment filter coefficient can be obtained.

  Note that the present invention can be realized in various modes. For example, it can be realized in a mode such as a video display device including a video processing device, a video processing method, and the like.

It is explanatory drawing which shows a trapezoid distortion correction | amendment notionally. It is explanatory drawing which shows the image | video displayed on the liquid crystal panel part 192 before and behind correction | amendment. It is a schematic diagram which shows notionally the method of pixel interpolation. 1 is a block diagram schematically showing a configuration of a projector as an embodiment of the present invention. 3 is a functional block diagram illustrating a configuration of a trapezoidal distortion correction unit 120. FIG. It is process drawing which shows typically the flow of the image data generation process after correction | amendment. It is explanatory drawing which shows the change of the coordinate of the image | video before and behind correction | amendment. It is a schematic diagram which shows notionally interpolation of a filter coefficient.

Next, embodiments of the present invention will be described in the following order based on examples.
A. First embodiment:
A-1. Overview of keystone correction:
A-2. Projector configuration:
A-3. Keystone distortion correction section:
A-4. Keystone distortion correction process:
A-5. Calculation of deformation rate:
A-6. Filter coefficient interpolation:
A-7. Adjustment of filter coefficient:
A-8. Effects of the embodiment:
B. Variations:

A. Example:
A-1. Overview of keystone correction:
The projector 100 as one embodiment of the present invention projects image light representing an image and displays the image on a projection surface such as a screen SC. The projector 100 is a projector capable of correcting a trapezoidal distortion of a video displayed on the screen SC and displaying a rectangular video when a rectangular video is input. Prior to the description of the configuration of the projector 100, an outline of trapezoidal distortion correction in the projector 100 of the present embodiment will be briefly described.

  FIG. 1 is an explanatory diagram conceptually showing trapezoidal distortion correction. As shown in the figure, when the projector 100 is arranged with an inclination in the horizontal direction (left-right direction) and the vertical direction (up-down direction) with respect to the screen SC, it is displayed on the liquid crystal panel unit 192. The video (pre-correction video IG0) is rectangular, whereas the video PIG0 projected on the screen SC has a trapezoidal distortion in each of the horizontal and vertical directions. In FIG. 1, for convenience of explanation, the liquid crystal panel unit 192 included in the projector 100 is displayed outside the projector 100.

  Therefore, when a projected image (corrected image IG1) distorted in the opposite direction to the image projected on the screen SC is formed on the liquid crystal panel unit 192 using a projective transformation technique, a rectangular image is displayed on the screen SC. PIG1 is displayed (FIG. 1). In this way, correction for making a video having a trapezoidal distortion appear rectangular (the shape of the video that should be originally displayed) is called a trapezoidal distortion correction. The uncorrected image IG0 in the present embodiment corresponds to the original image in the claims, and the corrected image IG1 corresponds to the modified image in the claims. Further, the trapezoidal distortion correction process in the present embodiment corresponds to the deformation process in the claims.

  FIG. 2 is an explanatory diagram showing images displayed on the liquid crystal panel unit 192 before and after correction. FIG. 2A shows the pre-correction video IG0, and FIG. 2B shows the post-correction video IG1. A broken line in FIG. 2B indicates the outer shape of the image IG0 before correction. Since the pre-correction video IG0 has been subjected to video processing so as to be displayed on the full frame of the liquid crystal panel unit 192, the broken line in FIG. 2B indicates the frame of the liquid crystal panel unit 192.

  As shown in FIG. 2B, in the corrected image IG1, the vicinity of the vertex (X1, Y1) has a high reduction rate (that is, the deformation rate is large) with respect to the pre-correction image IG0, and the vertex (X3, Y3). In the vicinity of, the reduction ratio relative to the pre-correction image IG0 is low (that is, the deformation ratio is small). That is, the corrected image IG1 has different deformation rates depending on the coordinates of the corrected image IG1. Here, the coordinates of the corrected image IG1 refer to pixel coordinates when the corrected image IG1 is displayed on the liquid crystal panel unit 192.

  As described above, since the deformation rate of the corrected image IG1 varies depending on the coordinates of the corrected image IG1, in the projector 100 of this embodiment, the corrected image IG1 is corrected for each pixel coordinate of the liquid crystal panel unit 192. A deformation rate for the image IG0 is calculated, and based on the calculated deformation rate, a pixel value of the corrected image IG1 is calculated to generate corrected image data representing the corrected image IG1 (described in detail later). Hereinafter, the pixel coordinates of the liquid crystal panel unit 192 when the corrected image IG1 is displayed are referred to as corrected coordinates. Of the pixel coordinates of the liquid crystal panel unit 192, the pixel coordinates of the area where the corrected image IG1 is not displayed are also referred to using the corrected coordinates. The coordinates obtained by converting the corrected coordinates to the coordinate positions (pixel coordinates of the liquid crystal panel) in the pre-correction video IG0 by reverse projection conversion are referred to as pre-correction coordinates.

  Since the pre-correction video IG0 and the post-correction video IG1 do not have an integer multiple correspondence, the calculated pre-correction coordinates include decimals. Therefore, in order to obtain the pixel value of the coordinate after correction, the pixel value of the coordinate before correction is estimated using the pixel value of 16 coordinates near the coordinate before correction. This is called pixel interpolation. That is, post-correction video data representing the post-correction video IG1 is created by performing pixel interpolation for every pixel on the pixel values of all pixels (coordinates) constituting the post-correction video IG1.

  FIG. 3 is a schematic diagram conceptually showing a pixel interpolation method. FIG. 3 illustrates a method of obtaining the pixel value of the uncorrected coordinate P0 (x, y) obtained by converting the corrected coordinate P1 (X, Y) by pixel interpolation. In the figure, the pre-correction coordinates P0 (x, y) are indicated by hatched circles, and the surrounding 16 coordinates are indicated by white circles. Since the pixel value of the pre-correction coordinate P0 (x, y) is obtained by pixel interpolation, the pixel value of the pre-correction coordinate P0 (x, y) is “interpolated pixel”, and the pixel values of the surrounding 16 coordinates are the pre-correction video data. Since it is known, it is also called “known pixel”.

  In FIG. 3, pixel values of 16 pixels that are known pixels are represented by DATA [m] [n] (m = 0, 1, 2, 3 (x direction); n = 0, 1, 2, 3 (y direction). ). The interpolation pixel is obtained by a convolution operation of the pixel value of 16 pixels and the filter coefficient. In this embodiment, when the pixel value of the interpolation pixel is obtained, the pixel value of the target pixel is obtained by performing horizontal interpolation after performing horizontal interpolation using 16 pixels around the target pixel. The filter coefficient is the distance between the interpolation pixel and the known pixel (for example, the distance between the known pixel of DATA [1] [1] and the interpolation pixel is dx in the horizontal direction (x direction) and dy in the vertical direction (y direction). This is a coefficient that takes into account the influence of the horizontal and horizontal pixels (distance in the x direction) between the known pixel and the interpolated pixel and the horizontal deformation rate (hereinafter also referred to as “horizontal deformation rate”). An interpolation filter coefficient is determined, and the vertical interpolation filter coefficient is determined based on a vertical distance between the known pixel and the interpolation pixel (distance in the y direction) and a vertical deformation rate (hereinafter also referred to as “vertical deformation rate”). It has been.

A-2. Projector configuration:
FIG. 4 is a block diagram schematically showing a configuration of a projector as an embodiment of the present invention. As illustrated, the projector 100 includes a video input unit 110, an IP conversion unit 112, a resolution conversion unit 114, a video synthesis unit 116, a trapezoidal distortion correction unit 120, a liquid crystal panel drive unit 140, and a frame buffer 150. A high-speed bus control unit 160, a low-speed bus control unit 162, a processor unit 170, an imaging unit 180, a sensor unit 182, an illumination optical system 190, a liquid crystal panel unit 192, and a projection optical system 194. It is structured in the center. Among the components described above, the components other than the illumination optical system 190, the liquid crystal panel unit 192, and the projection optical system 194 are connected to each other via the high-speed bus 102 or the low-speed bus 104.

  The video input unit 110 performs A / D conversion on an input video signal input from a DVD player or personal computer (not shown) via a cable, if necessary, and supplies the digital video signal to the IP conversion unit 112.

  The IP conversion unit 112 executes processing for converting the format of the video data supplied from the video input unit 110 from the interlace method to the progressive method, and supplies the obtained video data to the resolution conversion unit 114.

  The resolution conversion unit 114 performs a size enlargement process or a reduction process (that is, resolution conversion process) on the video data supplied from the IP conversion unit 112, and supplies the obtained video data to the video synthesis unit 116. To do.

  The video composition unit 116 synthesizes the video data supplied from the resolution conversion unit 114 and an OSD (On Screen Display) such as a menu screen, and writes the resultant data in the frame buffer 150 as pre-correction video data.

  The frame buffer 150 can store data of one frame or a plurality of frames. In this embodiment, an inexpensive and large-capacity DRAM (Dynamic Random Access Memory) is used as the frame buffer 150. The frame buffer 150 in this embodiment corresponds to a frame video storage unit in the claims.

  The trapezoidal distortion correction unit 120 corrects trapezoidal distortion that occurs when projection is performed with the projection axis of the projector 100 tilted with respect to the screen SC. Specifically, in order to display the pre-correction video represented by the pre-correction video data stored in the frame buffer 150 on the liquid crystal panel unit 192 in a shape that compensates for trapezoidal distortion, the correction processing is performed on the pre-correction video data. The corrected video data is supplied to the liquid crystal panel driving unit 140. The trapezoidal distortion correction unit 120 will be described in detail later.

  The liquid crystal panel drive unit 140 drives the liquid crystal panel unit 192 based on the digital video signal input through the trapezoidal distortion correction unit 120. The liquid crystal panel unit 192 includes a transmissive liquid crystal panel in which a plurality of pixels are arranged in a matrix. The liquid crystal panel unit 192 is driven by the liquid crystal panel driving unit 140, and changes the light transmittance in each pixel arranged in a matrix, thereby changing the illumination light emitted from the illumination optical system 190 to an effective image. An image for modulation into a new image light is formed. In this embodiment, the mode of the liquid crystal panel unit 192 is WUXGA, and the resolution is 1920 × 1200 dots. In this embodiment, the liquid crystal panel pixel coordinates are defined as x = 0 to 1919 and y = 0 to 1199. The liquid crystal panel unit 192 may have a resolution different from that of this embodiment.

  The illumination optical system 190 includes, for example, lamps such as a high pressure mercury lamp and an ultrahigh pressure mercury lamp, and other light emitters. The projection optical system 194 is attached to the front surface of the housing of the projector 100, expands the light modulated into the image light by the liquid crystal panel unit 192, and projects it onto the screen SC. The projection optical system 194 includes a zoom lens (not shown), and can change the degree of enlargement (zoom state) when projecting light transmitted through the liquid crystal panel unit 192. The liquid crystal panel driving unit 140, the liquid crystal panel unit 192, the illumination optical system 190, and the projection optical system 194 in this embodiment correspond to the image display unit in the claims.

  The processor unit 170 controls the operation of each unit in the projector 100 by reading and executing a control program stored in a storage unit (not shown). Further, corrected coordinates (X0 to X3, Y0 to Y3) described later based on the captured image captured by the imaging unit 180, the tilt of the projector 100 detected by the sensor unit 182 and the instruction from the user (see FIG. 2), a coordinate conversion coefficient (described in detail later) of the coordinate conversion matrix is calculated and output to the trapezoidal distortion correction unit 120. In this embodiment, the processor unit 170 is configured to receive an instruction from the user through an operation panel (not shown) provided in the main body of the projector 100. For example, the instruction from the user through a remote control is controlled by the remote controller. The processor unit 170 may receive the instruction from the user via the low-speed bus 104.

  The imaging unit 180 has a CCD camera and generates a captured video. The captured video generated by the imaging unit 180 is stored in a captured video memory (not shown). Note that the imaging unit 180 may have another imaging device instead of the CCD camera.

  The sensor unit 182 can detect the tilt angle formed by the CCD optical axis of the imaging unit 180 and the horizontal plane by detecting the tilt of the projector 100 from the vertical direction.

A-3. Keystone distortion correction section:
As described above, the trapezoidal distortion correction unit 120 generates post-correction video data in which the pre-correction video represented by the pre-correction video data stored in the frame buffer 150 is corrected to a shape that compensates for trapezoidal distortion. FIG. 5 is a functional block diagram illustrating a configuration of the trapezoidal distortion correction unit 120. The trapezoidal distortion correction unit 120 includes a cache block control unit 121, a cache block storage unit 122, a cache block tag information storage unit 123, an interpolation block reading unit 124, a pixel interpolation unit 125, a FIFO unit 126, and a register. Unit 127, control unit 128, coordinate conversion unit 129, deformation amount calculation unit 130, hit determination unit 131, filter coefficient selection unit 132, filter coefficient interpolation unit 133, interpolation result determination unit 134, filter The coefficient adjustment unit 135 and the filter coefficient storage unit 136 are mainly configured. In this embodiment, the filter coefficient interpolation unit 133 and the filter coefficient adjustment unit 135 in the trapezoidal distortion correction unit 120 perform fixed-point arithmetic.

  The cache block control unit 121 acquires the pre-correction video data stored in the frame buffer 150 for each pixel block composed of 8 × 8 pixels and stores it in the cache block storage unit 122. In the following description, a block of pixels composed of 8 × 8 pixels is referred to as a “cache block”. Further, the cache block control unit 121 updates the tag information stored in the cache block tag information storage unit 123. The cache block acquired by the cache block control unit 121 is specified by the hit determination unit 131 as described later.

  The cache block storage unit 122 can store a part of the pre-correction video data for one frame stored in the frame buffer 150 in units of cache blocks including 8 × 8 pixels. The cache block storage unit 122 includes 240 columns × 4 rows of block areas in which one cache block (video data composed of 8 × 8 pixels) can be stored. In this embodiment, the cache block storage unit 122 is composed of a small capacity and high speed SRAM (Static Random Access Memory).

  The cache block tag information storage unit 123 stores tag information that is management information when the cache block storage unit 122 is managed in units of block areas.

  The interpolation block reading unit 124 reads out an interpolation block composed of 4 × 4 pixels necessary for pixel interpolation in the pixel interpolation unit 125 from the cache block storage unit 122 and supplies it to the pixel interpolation unit 125.

  The pixel interpolation unit 125 performs pixel interpolation processing based on the interpolation block supplied from the interpolation block reading unit 124 and the adjusted filter coefficient (described later) supplied from the filter coefficient adjustment unit 135, and performs interpolation pixel ( The value of the corrected image pixel) is obtained and output to the liquid crystal panel driving unit 140 (FIG. 4) via the FIFO unit 126.

  The register unit 127 stores parameters supplied from the processor unit 170. Specifically, the register unit 127 includes parameters such as the frame width and frame height of the pre-correction video, the coordinate conversion coefficients A to H of the coordinate conversion matrix, and background colors (for example, black, gray, blue). Etc.) is stored as background color data. The coordinate conversion coefficients A to H are calculated by the processor unit 170 using the following determinant (formula 1) of projective transformation. Specifically, the processor unit 170 converts the uncorrected coordinates (x0 to x3, y0 to y3) (see FIG. 2) into the corrected coordinates (X0 to X3, Y0 to Y3) (see FIG. 2) by projective transformation. As converted, the four coordinates (X0 to X3, Y0 to Y3) of the corrected image IG1 are input to the determinant (formula 1) to derive the coordinate conversion coefficients A to H, and the register unit 127 To store.

  In this embodiment, before the correction image data generation process is started, the image PIG0 displayed on the screen SC before the trapezoidal distortion correction is imaged by the imaging unit 180. The processor unit 170 (FIG. 4) obtains the coordinates (X0 to X3, Y0 to Y3) (FIG. 2B) of the four vertices in the corrected image IG1 after the trapezoidal distortion correction based on the captured image. Yes.

  Note that the sensor unit 182 may detect the inclination of the projector 100 from the vertical direction, and obtain corrected coordinates (X0 to X3, Y0 to Y3) based on the detected angle. Further, the keystone distortion correction may be manually performed by the user operating a key such as a remote controller. In such a case, the processor unit 170 obtains corrected coordinates (X0 to X3, Y0 to Y3) based on an instruction from the user received via the remote control unit.

  The control unit 128 controls the trapezoidal distortion correction unit 120 in general. For example, a frame start signal indicating the start of a frame is output to the coordinate conversion unit 129 in accordance with the synchronization signal input to the control unit 128. For example, when 60 frames are displayed per second, the synchronization signal is input every 1/60 seconds.

  The coordinate conversion unit 129 uses the coordinate conversion coefficients A to H supplied from the register unit 127 and the following (Expression 2) and (Expression 3) to correct the coordinates of the corrected image IG1 after performing the trapezoidal distortion correction. The value (post-correction coordinates) is converted into the coordinate value (pre-correction coordinates) of the pre-correction video IG0 (rectangular video), and the pre-correction coordinates are supplied to the deformation amount calculation unit 130. Since the pre-correction video IG0 and the post-correction video IG1 do not have an integer multiple correspondence, the pre-correction coordinates calculated by the coordinate conversion unit 129 include decimals.

  Here, the above-described calculation method of the pre-correction coordinates will be described. Since the corrected image IG1 is considered to be an image obtained by projective transformation of the uncorrected image IG0, the uncorrected coordinates are based on the following (Expression 2) and (Expression 3) with respect to the corrected coordinates. Then, it is calculated by reverse projective transformation. It is assumed that the uncorrected coordinates (x, y) are converted to corrected coordinates (X, Y) by projective transformation.

上記（式２）、（式３）中の係数Ａ〜Ｈは、レジスター部１２７に記憶されている。

The coefficients A to H in the above (Expression 2) and (Expression 3) are stored in the register unit 127.

  The deformation amount calculation unit 130 divides the uncorrected coordinates supplied from the coordinate conversion unit 129 into an integer part and a decimal part, supplies the integer part to the hit determination unit 131, and supplies the decimal part to the filter coefficient selection unit 132. To do. Further, the deformation amount calculation unit 130 calculates the deformation amount of the corrected image IG1 with respect to the pre-correction image IG0 based on the pre-correction coordinates supplied from the coordinate conversion unit 129, and supplies it to the filter coefficient selection unit 132. A method of calculating the deformation amount will be described in detail later.

  The filter coefficient selection unit 132 calculates the deformation rate based on the deformation amount supplied from the deformation amount calculation unit 130. The filter coefficient selection unit 132 selects, based on the calculated deformation rate and the decimal part supplied from the deformation amount calculation unit 130, a filter coefficient used when performing pixel interpolation processing from the filter coefficient table, The selected filter coefficient is supplied to the filter coefficient interpolation unit 133. The filter coefficient table is a table showing the relationship between the distance between the interpolation pixel and the known pixel, the deformation rate, and the filter coefficient shown in FIG. 3, and the result calculated in advance is stored in the filter coefficient storage unit 136. . The filter coefficient storage unit 136 stores a filter coefficient table in a fixed-point format. In this embodiment, a table is prepared for each of the horizontal filter coefficient and the vertical filter coefficient. The horizontal filter coefficient and the vertical filter coefficient are collectively referred to simply as “filter coefficient”.

  In this embodiment, the deformation rate is 1, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 10 in total. Discretized into pieces. Therefore, when the deformation rate calculated by the filter coefficient selection unit 132 does not correspond to the discretized value, it corresponds to two deformation rates before and after the calculated deformation rate for the interpolation calculation. Select the filter coefficient. For example, when the deformation rate is 0.26, filter coefficients corresponding to the deformation rate 0.2 and the deformation rate 0.3 are selected.

  In the present embodiment, as described above, the pixel value of the target pixel is obtained by performing horizontal interpolation after performing horizontal interpolation using 16 pixels around the interpolation pixel (FIG. 3). Therefore, first, based on the horizontal distance between the known pixel and the interpolation pixel and the horizontal deformation rate, the horizontal filter coefficient corresponding to each of the four known pixels having the same vertical position is selected.

  Specifically, the horizontal filter coefficients corresponding to DATA [0] [0], DATA [1] [0], DATA [2] [0], DATA [3] [0] are based on the horizontal deformation rate. Select two each. The horizontal filter coefficients corresponding to DATA [0] [1], DATA [0] [2], DATA [0] [3] are the same as DATA [0] [0], DATA [1] [1], DATA The horizontal filter coefficients corresponding to [1] [2], DATA [1] [3] are the same as DATA [1] [0], DATA [2] [1], DATA [2] [2], DATA [ The horizontal filter coefficients corresponding to 2] [3] are the same as DATA [2] [0], and the horizontal filter coefficients corresponding to DATA [3] [1], DATA [3] [2], DATA [3] [3] The filter coefficient is the same as DATA [3] [0].

  Subsequently, based on the vertical distance between the known pixel and the interpolated pixel and the vertical deformation rate, the vertical filter coefficient corresponding to each of the four known pixels having the same horizontal position is selected. This vertical filter coefficient is used when vertical interpolation is performed using four pixels having the same x coordinate as P0 (x, y) obtained by horizontal interpolation. Specifically, the vertical filter coefficients corresponding to DATA [0] [0], DATA [0] [1], DATA [0] [2], and DATA [0] [3] are based on the vertical deformation rate. , Select two each.

  The filter coefficient interpolation unit 133 performs interpolation for each of the horizontal filter coefficient and the vertical filter coefficient selected by the filter coefficient selection unit 132, and calculates an interpolation horizontal filter coefficient and an interpolation vertical filter coefficient. The interpolated horizontal filter coefficient and the interpolated vertical filter coefficient are collectively referred to simply as “interpolated filter coefficient”. In the present embodiment, the filter coefficient corresponding to the deformation ratio calculated by the filter coefficient selection unit 132 is calculated by linear interpolation, but may be obtained by other interpolation methods such as spline interpolation and Lagrange interpolation.

  The interpolation result determination unit 134 determines whether or not the sum of the interpolation horizontal filter coefficient and the interpolation vertical filter coefficient obtained by the interpolation in the filter coefficient interpolation unit 133 is a reference value. In this embodiment, the reference value = 1, but it may not be 1. For example, it may be set to a value smaller than 1 such as a reference value = 0.95. In this way, the gradation of the interpolation pixel can be lowered. Alternatively, the reference value may be set to a value larger than 1, such as 1.05. In this way, the gradation of the interpolation pixel can be increased.

  The filter coefficient adjustment unit 135 sums the interpolation horizontal filter coefficient and the interpolation vertical filter coefficient when the sum of the interpolation horizontal filter coefficient and the interpolation vertical filter coefficient is not the reference value (= 1) in the interpolation result determination unit 134. Adjust so that becomes the reference value. A specific adjustment method will be described later.

  Whether the hit determination unit 131 stores the pixel value of the coordinates used for pixel interpolation in the pixel interpolation unit 125 in the cache block storage unit 122 based on the integer part of the coordinates before correction supplied from the deformation amount calculation unit 130. Determine whether or not. Hereinafter, this determination is referred to as “hit determination”. As a result of the hit determination, when a pixel value necessary for pixel interpolation is not stored in the cache block storage unit 122, a request for acquiring a necessary cache block is issued to the cache block control unit 121. As a result of the hit determination, when a pixel value necessary for pixel interpolation is stored in the cache block storage unit 122, the read position in the cache block storage unit 122 is supplied to the interpolation block reading unit 124.

A-4. Keystone distortion correction process:
FIG. 6 is a process diagram schematically showing the flow of the corrected video data generation process. In the trapezoidal distortion correction unit 120 in this embodiment, when a frame start signal is output from the control unit 128 (FIG. 5) and input to the coordinate conversion unit 129 (FIG. 5), the corrected video data generation process is started. .

  The coordinate conversion unit 129 (FIG. 5) determines whether or not a frame start signal has been input (step S102). The coordinate conversion unit 129 waits until the frame start signal is input (NO in step S102). When the frame start signal is input (YES in step S102), as described above, the coordinates of the corrected video are obtained. The coordinates before correction are obtained by converting the coordinates after correction (step S104).

  When the pre-correction coordinates calculated by the coordinate conversion unit 129 are supplied to the deformation amount calculation unit 130 (FIG. 5), the deformation amount calculation unit 130 calculates the deformation rate of the corrected image with respect to the pre-correction image at the pre-correction coordinates. (Step S106).

  When the integer part of the coordinates before correction is supplied from the deformation amount calculation unit 130 to the hit determination unit 131 (FIG. 5), the hit determination unit 131 performs hit determination and determines that the target cache block is stored. The interpolation block readout position is supplied to the interpolation block readout unit 124. When the interpolation block readout position is supplied to the interpolation block readout unit 124, the interpolation block readout unit 124 (FIG. 5) reads out the interpolation block from the cache block storage unit 122 based on the supplied readout position (step S108). .

  On the other hand, the filter coefficient selection unit 132 (FIG. 5) selects the filter coefficient from the filter coefficient table as described above based on the decimal part of the coordinates before correction supplied from the deformation amount calculation unit 130 and the deformation amount. (Step S110). When the filter coefficient is supplied from the filter coefficient selection unit 132, the filter coefficient interpolation unit 133 (FIG. 5) performs the interpolation of the filter coefficient, and the interpolation filter coefficient obtained by the interpolation is determined as the interpolation result determination unit 134 (FIG. 5). 5) (step S112).

  The interpolation result determination unit 134 determines whether or not the sum of the interpolation horizontal filter coefficient and the interpolation vertical filter coefficient supplied from the filter coefficient interpolation unit 133 is 1, and the determination result together with the interpolation filter coefficient is a filter coefficient adjustment unit 135. (FIG. 5) is supplied (step S114). When the determination result supplied from the interpolation result determination unit 134 indicates that the sum of the filter coefficients is not 1, the filter coefficient adjustment unit 135 adjusts the filter coefficient based on the supplied filter coefficient, and the adjustment filter The coefficient is supplied to the pixel interpolation unit 125 (FIG. 5) (step S116).

  Hit determination unit 131, cache block control unit 121, cache block storage unit 122, cache block tag information storage unit 123, interpolation block reading processing in interpolation block reading unit 124, filter coefficient selection unit 132, filter coefficient interpolation unit 133, The filter coefficient calculation process in the interpolation result determination unit 134 and the filter coefficient adjustment unit 135 is performed at the same time. When both processes are completed, the process proceeds to step S118.

  When the interpolation block is supplied from the interpolation block reading unit 124 and the adjustment filter coefficient is supplied from the filter coefficient adjustment unit 135, the pixel interpolation unit 125 (FIG. 5) uses the supplied interpolation block and adjustment filter coefficient. Then, the pixel value of the corrected coordinates is calculated by the convolution operation (step S118).

  By repeating the above steps S102 to S118 from the corrected coordinates (X, Y) = (0, 0) to (frame width −1, frame height −1), corrected video data is generated. The In this embodiment, since the frame width is 1920 and the frame height is 1200, steps S102 to S116 are repeated until the corrected coordinates (X, Y) = (0, 0) to (1919, 1199). When the frame width and the frame height are different from those of the present embodiment, the steps S102 to S116 are performed after the corrected coordinates (X, Y) = (0, 0) to (frame width−1, frame height− By repeating the process up to 1), corrected video data is generated.

  In addition, when a plurality of videos are continuously input (for example, when a moving image is input), the above processing may be repeated for each video to perform pixel interpolation, or the first As described above, the adjustment filter coefficient is calculated for each coordinate for the image, the calculated adjustment filter coefficient is stored in the register unit 127, and the second and subsequent images are corrected for trapezoidal distortion. Pixel interpolation may be performed using the adjustment filter coefficient stored in 127. The former case is suitable for processing by hardware. In the latter case, the hardware configuration can be omitted by processing with software.

A-5. Calculation of deformation rate:
The deformation amount calculation unit 130 (FIG. 5) converts, for each coordinate of the corrected image, target coordinates for calculating a pixel value and its surrounding eight coordinates into the coordinates of the uncorrected image by reverse projection transformation. Find the amount of deformation of coordinates.

  FIG. 7 is an explanatory diagram showing changes in the coordinates of the image before and after correction. FIG. 7A shows the pre-correction video IG0, and FIG. 7B shows the post-correction video IG1. The coordinates before correction shown in FIG. 7B, which are the results of converting the coordinates (a_x, a_y) in the corrected image IG1 and the surrounding eight coordinates into the coordinates of the image IG0 before correction, are shown in FIG. It shows. Hereinafter, the target coordinates for obtaining the deformation amount are also referred to as “target corrected coordinates”.

  In this embodiment, as shown in FIG. 7B, the post-target correction coordinates of the post-correction video IG1 are (a_x, a_y), and the peripheral pixel coordinates are the coordinates of the eight pixels surrounding the target post-correction coordinates. (A_x-1, a_y-1), (a_x + 0, a_y-1), (a_x + 1, a_y-1), (a_x-1, a_y + 0), (a_x + 1, a_y + 0), (a_x-1, a_y + 1), (a_x + 0) , A_y + 1), (a_x + 1, a_y + 1). The pre-correction coordinates are calculated by the reverse projection transformation of the above-described projection transformation.

  The coordinates before correction calculated by the inverse projective transformation are represented by src (0,0), src (1,0), src (2,0), src (0,1), src (1) shown in FIG. , 1), src (2, 1), src (0, 2), src (1, 2), src (2, 2). That is, the coordinates after correction (a_x−1, a_y−1) → the coordinates before correction src (0,0), the coordinates after correction (a_x + 0, a_y−1) → the coordinates before correction src (1,0), the coordinates after correction ( a_x + 1, a_y-1) → pre-correction coordinates src (2,0), post-correction coordinates (a_x−1, a_y + 0) → pre-correction coordinates src (0,1), post-correction coordinates (a_x, a_y) → pre-correction coordinates src (1, 1), corrected coordinates (a_x + 1, a_y + 0) → corrected coordinates src (2,1), corrected coordinates (a_x−1, a_y + 1) → corrected coordinates src (0,2), corrected coordinates (A_x + 0, a_y + 1) → pre-correction coordinates src (1,2), post-correction coordinates (a_x + 1, a_y + 1) → pre-correction coordinates src (2,2).

  The deformation amount calculating unit 130 calculates the horizontal distance between adjacent pre-correction coordinates shown in FIG. Specifically, the horizontal distance between src (0,0) and src (1,0), the horizontal distance between src (1,0) and src (2,0),..., Src (1, The horizontal distance between 2) and src (2, 2) is calculated, the maximum value is obtained, and the maximum value is set as the horizontal deformation amount. As shown in FIG. 7B, since the horizontal distance between adjacent corrected coordinates is “1”, the filter coefficient selection unit 132 sets the reciprocal of the horizontal deformation amount as the horizontal deformation rate. In the present embodiment, the deformation rate is calculated using the maximum horizontal distance between adjacent uncorrected coordinates, but an average value of adjacent horizontal distances between uncorrected coordinates may be used.

  Similarly, the vertical distance between adjacent uncorrected coordinates is calculated to determine the maximum value, and the maximum value is used as the vertical deformation amount. The filter coefficient selection unit 132 sets the reciprocal of the vertical deformation amount as the vertical deformation rate.

  For example, when the coordinates near the upper right vertex of the corrected image IG1 shown in FIG. 7B are the corrected coordinates (a_x, a_y), as shown in FIG. The maximum value of the horizontal distance (horizontal deformation amount) = 2, and the horizontal deformation rate = 1/2. Further, as shown in FIG. 7A, the maximum value of the vertical distance between adjacent uncorrected coordinates (vertical deformation amount) = 2.3, and the vertical deformation ratio = 1 / (2.3).

A-6. Filter coefficient interpolation:
FIG. 8 is a schematic diagram conceptually showing interpolation of filter coefficients. FIG. 8 illustrates horizontal filter coefficients corresponding to the known pixels DATA [0] [0], DATA [1] [0], DATA [2] [0], and DATA [3] [0] shown in FIG. ing. The numbers described in the circles and squares in the figure are the x values of the known pixel DATA [x] [y], the circles are the horizontal filter coefficients in the case of the horizontal deformation rate α, and the squares are the horizontal deformation rate β. Shows the horizontal filter coefficient. The interpolation kernel in the case of the horizontal deformation rate α is indicated by a solid line, and the interpolation kernel in the case of the horizontal deformation rate β is indicated by a broken line. In the following description, the horizontal deformation rate is simply referred to as “deformation rate”, and the horizontal filter coefficient is simply referred to as “filter coefficient”.

  When the deformation rate of the coordinate after object correction = sf (α <sf <β), by performing linear interpolation using the filter coefficient when the deformation rate = α and the deformation rate = β, the case where the deformation rate = sf The filter coefficient can be calculated. For example, when the filter coefficient (interpolation filter coefficient) corresponding to the known pixel DATA [0] [0] when the deformation rate = sf is NewCoef [0], NewCoef [0] is obtained by the following (formula 4). Can do. Note that NewCoef [1], NewCoef [2], and NewCoef [3] can be similarly obtained (FIG. 8).

NewCoef [0] = R × (filter coefficient corresponding to DATA [0] [0] when deformation rate = β) + (1−R) × (DATA [0] [0] when deformation rate = α) Filter coefficient corresponding to (...) (Equation 4)
Where R = (sf−α) / (β−α)

  Hereinafter, an example will be described with specific numerical values. When sf = 0.125, the filter coefficient selection unit 132 selects each filter coefficient when the deformation ratio = 0.1, 0.2. That is, α = 0.1 and β = 0.2. The filter coefficient in the case of the deformation rate α is Coefα [x] (x = 0, 1, 2, 3), and the filter coefficient in the case of the deformation rate β is Coefβ [x] (x = 0, 1, 2, 3). Assuming that the filter coefficient in the case of the deformation rate sf is NewCoef [x] (x = 0, 1, 2, 3), the ideal value of each filter coefficient is as follows.

Coefα [0] = − 0.220
Coefα [1] = 0.682
Coefα [2] = 0.838
Coefα [3] = − 0.300
Coefα Total = 1.000

Coefβ [0] = − 0.111
Coefβ [1] = 0.575
Coefβ [2] = 0.713
Coefβ [3] = − 0.177
Coefβ Total = 1.000

NewCoef [0] = − 0.19275
NewCoef [1] = 0.65525
NewCoef [2] = 0.080675
NewCoef [3] = − 0.26925
NewCoef Total = 1.000

  As described above, ideally, the sum of the filter coefficients at α = 0.1 and β = 0.2 is 1, and as a result, the sum of the filter coefficients at sf = 0.125 is also obtained. It is 1. However, the filter coefficient storage unit 136 stores the filter coefficient in a fixed point format. Therefore, the filter coefficients stored in the filter coefficient storage unit 136 (for example, the filter coefficients at α = 0.1 and β = 0.2) often do not have the ideal values described above. For this reason, the sum of the filter coefficients at α = 0.1 and β = 0.2 often does not become one. In addition, since the filter coefficient interpolation unit 133 performs fixed-point arithmetic, the sum of the filter coefficients at sf = 0.125 obtained using the filter coefficient selected from the filter coefficient storage unit 136 is often not 1. Below, each filter coefficient in the case of sign 1bit, integer 1bit, fraction 6bit is illustrated. Note that minus is expressed in two's complement.

Coefα [0] = 11110010 (−0.218750)
Coefα [1] = 00101100 (0.687500)
Coefα [2] = 001110110 (0.843750)
Coefα [3] = 11101101 (−0.296875)
Coefα Total = 1.015625

Coefβ [0] = 11111001 (−0.109375)
Coefβ [1] = 001010101 (0.578125)
Coefβ [2] = 001011110 (0.718750)
Coefβ [3] = 11110101 (−0.171875)
Coefβ Total = 1.015625

When the interpolation filter coefficient at sf = 0.125 is obtained by interpolation using the above filter coefficients, the following is obtained.
NewCoef [0] = 11110100 (−0.187500)
NewCoef [1] = 00101010 (0.656250)
NewCoef [2] = 001110100 (0.812500)
NewCoef [3] = 11101111 (−0.265625)
NewCoef Total = 1.015625

  As described above, since the filter coefficient storage unit 136 of the present embodiment stores the filter coefficient table in the fixed-point format, the sum of the filter coefficients stored in the filter coefficient storage unit 136 does not become 1, and the filter coefficient Since the interpolation unit 133 performs a fixed-point operation, an operation error occurs, and the sum of interpolation filter coefficients calculated by interpolation using the filter coefficients stored in the filter coefficient storage unit 136 is often not 1. . Therefore, in the present embodiment, as shown below, the filter coefficient is adjusted using the interpolation filter coefficient.

A-7. Adjustment of filter coefficient:
In the interpolation result determination unit 134 (FIG. 5), when the sum of the interpolation filter coefficients supplied from the filter coefficient interpolation unit 133 is not 1, the filter coefficient adjustment unit 135 (FIG. 5) sets the sum of the filter coefficients to 1. Adjust the filter coefficient and calculate the adjusted filter coefficient. Specifically, the filter coefficient adjustment unit 135 divides the difference between the sum of the interpolation filter coefficients supplied from the interpolation result determination unit 134 and 1 (reference value) by 2, and obtains from NewCoef [0] and NewCoef [3]. Add and subtract. The reason for adjusting in this way is that NewCoef [0] and NewCoef [3] are farther away from the interpolated pixel than NewCoef [1] and NewCoef [2], and therefore the influence of the filter coefficient is small. is there.

  The adjustment of the filter coefficient will be described using the interpolation filter coefficients NewCoef [0] to NewCoef [3] described in the section A-6. As described above, the total sum of NewCoef [0] to NewCoef [3] is 1.015625 (= 01000001). Therefore, the difference value 0.015625 from 1 is divided by 2, and subtracted from NewCoef [0] and NewCoef [3]. When the adjustment filter coefficient is NewCoefA [0] to NewCoefA [3], the following is obtained.

NewCoefA [0] = NewCoef [0]-(0.0156625 / 2)
= 11110100 (−0.187500) − (0.0156625 / 2)
NewCoefA [1] = NewCoef [1]
= 00101010 (0.656250)
NewCoefA [2] = NewCoef [2]
= 00110100 (0.812500)
NewCoefA [3] = NewCoef [3]-(0.0156625 / 2)
= 11101111 (-0.265625)-(0.0156625 / 2)

  As described above, in this embodiment, each circuit in the trapezoidal distortion correction unit 120 performs a fixed-point operation, and the decimal part is represented by 6 bits. Therefore, (0.0156625 / 2) cannot be represented. As described above, when the result obtained by dividing the difference value by 2 cannot be represented by the decimal part 6 bits, the difference value is subtracted from NewCoef [0]. Specifically, it is as follows.

NewCoefA [0] = NewCoef [0]-(0.015625)
= 11110100 (−0.187500) −00000001 (0.015625)
= 11110011 (−0.203125)
NewCoefA [1] = NewCoef [1]
= 00101010 (0.656250)
NewCoefA [2] = NewCoef [2]
= 00110100 (0.812500)
NewCoefA [3] = NewCoef [3]
= 11101111 (-0.265625)
NewCoefA Total = 1.0

  In this way, the filter coefficient adjustment unit 135 basically divides the difference between the sum of the interpolation filter coefficients supplied from the interpolation result determination unit 134 and 1 (reference value) by 2, and NewCoef [0] Addition and subtraction to NewCoef [3]. If the result of dividing the difference value between the sum of the interpolation filter coefficients and 1 (reference value) by 2 cannot be expressed by the decimal part 6 bits, the difference value is added to or subtracted from NewCoef [0]. Note that NewCoef [3] may be added or subtracted. The vertical filter coefficient is similarly adjusted.

A-8. Effects of the embodiment:
As described above, according to the projector 100 in this embodiment, the calculation error of the interpolation filter coefficient obtained by the interpolation in the filter coefficient interpolation unit 133 is corrected in the filter coefficient adjustment unit 135, so that the sum of the filter coefficients Thus, an appropriate filter coefficient (adjusted filter coefficient) can be obtained. When pixel interpolation is performed using the interpolation filter coefficient calculated by the filter coefficient interpolation unit 133, the gradation of the image after the trapezoidal distortion correction processing is higher than the gradation of the image before the correction due to the calculation error due to the fixed point calculation, In some cases, it may not be possible to obtain an image equivalent to the uncorrected image. On the other hand, according to the projector 100 in this embodiment, it is possible to perform appropriate pixel interpolation as compared with the case where the filter coefficient is not adjusted, and the image quality of the image after the trapezoidal distortion correction processing is improved. Deterioration can be suppressed.

  Further, by using a fixed-point format in the filter coefficient interpolation unit 133, the filter coefficient adjustment unit 135, and the filter coefficient storage unit 136, hardware resources can be reduced. Further, by adjusting the filter coefficient, it is possible to perform an appropriate trapezoidal distortion correction process with an appropriate filter coefficient even if a floating point arithmetic unit that requires a large amount of hardware resources is not provided.

  In addition, by performing arithmetic processing by fixed-point arithmetic, it is possible to speed up the processing by hardware, and it is possible to calculate the filter coefficient every time the trapezoidal distortion correction processing is performed for one video (that is, Real-time processing is possible). Therefore, for example, when displaying a continuous image with the tilt of the projector 100 changing at any time, the trapezoidal distortion correction can be performed following the change in the tilt of the projector 100.

  Further, when the filter coefficient adjustment unit 135 adjusts the filter coefficient, the image can be easily corrected by changing the reference value, such as raising or lowering the gradation.

B. Modifications The present invention is not limited to the above-described embodiments, and can be implemented in various modes without departing from the scope of the invention.

  (1) In the present embodiment described above, an example was shown in which the difference value between the sum of the interpolation filter coefficients and the reference value (= 1) is divided by 2 and added to or subtracted from New Coef [0] and New Coef [3]. However, the filter coefficient may be adjusted by other methods. For example, the difference value between the sum of interpolation filter coefficients and the reference value (= 1) may be divided by 2 and added to or subtracted from New Coef [1] and New Coef [2]. As described above, since the filter coefficient in the known pixel that is close to the interpolation pixel has a great influence on the pixel interpolation, the image after the trapezoidal distortion correction tends to be sharp.

  (2) Alternatively, the difference value between the sum of the interpolation filter coefficients and the reference value (= 1) may be divided by 4, and addition / subtraction may be performed on all of New Coef [0] to New Coef [3].

  (3) In the modified example (1), when the difference value is divided by 2 and cannot be expressed by the decimal part 6 bits, the difference value is added to or subtracted from either New Coef [1] or New Coef [2]. You may do it. In the above-described embodiment and modification example (1), when the difference value is divided by 2, if it cannot be expressed by the decimal part 6 bits, one of the two coefficients for allocating the difference value is 2 You may give the remainder which was not divisible.

  (4) The number of pixels in one frame, the number of pixels in the cache block, and the number of pixels in the interpolation block (the number of pixels used for pixel interpolation) are not limited to the above embodiment. When the number of pixels used for pixel interpolation is different from that in the above embodiment, it is preferable to appropriately adjust the filter coefficient so that the sum of the filter coefficients becomes the reference value.

  (5) In the embodiment described above, the projector 100 modulates the light from the illumination optical system 190 using the transmissive liquid crystal panel unit 192, but is not limited to the transmissive liquid crystal panel unit 192. For example, using a digital micromirror device (DMD (registered trademark): Digital Micro-Mirror Device), a reflective liquid crystal panel (LCOS (registered trademark): Liquid Crystal on Silicon), etc., from the illumination optical system 190 The light may be modulated. Further, it may be a CRT projector that projects an image on a small CRT (cathode ray tube) onto a projection surface, or may be configured to generate image light using an organic EL.

  (6) In the above embodiment, a part of the function realized by hardware may be realized by software, or a part of the function realized by software may be realized by hardware.

  (7) In the above embodiment, the trapezoidal distortion correction unit 120 calculates the deformation rate for each pixel (coordinate) of the corrected coordinates using a plurality of surrounding pixels. Is not limited to the embodiments described above. For example, one pixel (coordinate), the neighboring pixel (coordinate) in the horizontal and vertical directions, and the deformation rate are calculated separately, and the deformed video is divided into blocks for each pixel (coordinate) and deformed for each block The rate may be derived, and the deformation rate corresponding to the block may be applied to all the pixels (coordinates) included in the block.

  (8) In the above-described embodiment, when the deformation rate in the corrected coordinates is derived, the deformation rate is derived using 8 coordinates around the target corrected coordinates. It is not limited to the above-described embodiment. For example, the deformation rate may be derived using one coordinate (one coordinate each in the x direction and the y direction) adjacent to the target corrected coordinate.

  (9) In the above embodiment, the example in which the pixel value of the target pixel is obtained by performing horizontal interpolation after performing horizontal interpolation using 16 pixels around the target pixel is shown. ) And the interpolation distance in the vertical direction (y direction) may be used to perform pixel interpolation using a two-dimensional interpolation filter.

  (10) In the above-described embodiments, the projector is exemplified as the video display device. However, the video display device is not limited to the projection type projector, but a liquid crystal panel, an organic EL (Electro-Luminescence) panel, or the like. It may be a video display device that displays video on the video display unit. For example, when a rectangular image is displayed on an image display unit configured by forming a thin organic EL display panel into a cylindrical shape, it is necessary to perform a predetermined deformation process on the input image. In addition, for example, deformation processing that causes an image with an aspect ratio different from the aspect ratio of the original image, or deformation processing that has various shapes such as a parallelogram or other quadrangle may be performed.

  Further, for example, in a digital camera (video display device) including a liquid crystal panel (video display unit), the video processing device may perform trapezoidal distortion correction. In this case, the video processing apparatus corrects distortion (perspective distortion) that occurs when the camera sensor is not parallel to the subject and outputs the corrected image to the video display unit. An image shot so as to be parallel is displayed on the image display unit. Also in this case, it is possible to suppress the deterioration of the image quality by deriving the filter coefficients based on the deformation rate and adjusting the sum of the filter coefficients to be the reference value.

  (11) In the above embodiment, the case where the calculation error caused by performing the fixed point calculation is corrected has been described as an example. However, the present invention is not limited to the calculation error caused by performing the fixed point calculation. The present invention can be applied to an error caused by various reasons such as an error caused by nonlinear interpolation and an error caused when different interpolation is performed between horizontal interpolation and vertical interpolation.

DESCRIPTION OF SYMBOLS 100 ... Projector 102 ... High speed bus 104 ... Low speed bus 110 ... Video input part 112 ... IP conversion part 114 ... Resolution conversion part 116 ... Video composition part 120 ... Trapezoid distortion correction part 121 ... Cache block control part 122 ... Cache block memory | storage part 123 ... tag information storage unit for cache block 124 ... interpolation block reading unit 125 ... pixel interpolation unit 126 ... FIFO unit 127 ... register unit 128 ... control unit 129 ... coordinate conversion unit 130 ... deformation amount calculation unit 131 ... hit determination unit 132 ... filter Coefficient selection section 133 ... Filter coefficient interpolation section 134 ... Interpolation result determination section 135 ... Filter coefficient adjustment section 136 ... Filter coefficient storage section 140 ... Liquid crystal panel drive section 150 ... Frame buffer 160 ... High-speed bus control section 162 ... Low-speed bus control section 170 ... Processor part 180 ... Imaging part 182 ... Sensor part 190 ... Illumination optical system 192 ... Liquid crystal panel part 194 ... Projection optical system SC ... Screen IG0 ... Image before correction IG1 ... Image after correction

Claims (6)

A video processing device,
A deformation rate deriving unit for deriving a deformation rate of the post-deformation video with respect to the original video for pixels in the post-deformation video when the original video is subjected to deformation processing based on a predetermined rule;
A storage unit for storing a filter coefficient for deriving a pixel value of the post-deformation video by pixel interpolation using a pixel value of the original video, and a plurality of the filters for each predetermined discrete deformation rate A filter coefficient storage unit that associates and stores coefficients;
A filter coefficient selection unit that selects the plurality of filter coefficients based on the deformation rate for each pixel of the post-deformation image, wherein the derived deformation rate is stored in the filter coefficient storage unit. A filter coefficient selecting unit that selects filter coefficients corresponding to a plurality of deformation ratios close to the derived deformation ratio,
In the filter coefficient selection unit, when filter coefficients corresponding to a plurality of deformation ratios are selected, the derived deformation ratio is obtained by performing interpolation based on the filter coefficients selected corresponding to the plurality of deformation ratios. A filter coefficient interpolation unit that generates a filter coefficient corresponding to
An interpolation result determination unit that determines whether or not an interpolation filter coefficient that is a filter coefficient derived in the filter coefficient interpolation unit satisfies a predetermined condition;
In the interpolation result determination unit, when it is determined that the predetermined condition is not satisfied, the interpolation filter coefficient is adjusted and a filter coefficient adjustment unit that generates an adjustment filter coefficient and the adjustment filter coefficient A pixel interpolation unit that derives and outputs a pixel value of each pixel of the transformed video by interpolation;
A video processing apparatus. The video processing apparatus according to claim 1,
The image processing apparatus, wherein the deformation process is a trapezoidal distortion correction process. The video processing apparatus according to claim 1 or 2,
The predetermined condition is:
A video processing apparatus in which a sum of a plurality of the interpolation filter coefficients corresponding to the derived deformation rate is 1. The video processing device according to any one of claims 1 to 3,
The image processing apparatus, wherein the deformation rate deriving unit derives the deformation rate for each pixel of the image after deformation. A video display device,
A video processing device;
A video display unit for displaying a video corresponding to the video data based on the video data output from the video processing device;
With
The video processing device includes:
A deformation rate deriving unit for deriving a deformation rate of the post-deformation video with respect to the original video for pixels in the post-deformation video when the original video is subjected to deformation processing based on a predetermined rule;
A storage unit for storing a filter coefficient for deriving a pixel value of the post-deformation video by pixel interpolation using a pixel value of the original video, and a plurality of the filters for each predetermined discrete deformation rate A filter coefficient storage unit that associates and stores coefficients;
A filter coefficient selection unit that selects the plurality of filter coefficients based on the deformation rate for each pixel of the post-deformation image, wherein the derived deformation rate is stored in the filter coefficient storage unit. A filter coefficient selecting unit that selects filter coefficients corresponding to a plurality of deformation ratios close to the derived deformation ratio,
In the filter coefficient selection unit, when filter coefficients corresponding to a plurality of deformation ratios are selected, the derived deformation ratio is obtained by performing interpolation based on the filter coefficients selected corresponding to the plurality of deformation ratios. A filter coefficient interpolation unit that generates a filter coefficient corresponding to
An interpolation result determination unit that determines whether or not an interpolation filter coefficient that is a filter coefficient derived in the filter coefficient interpolation unit satisfies a predetermined condition;
In the interpolation result determination unit, when it is determined that the predetermined condition is not satisfied, the interpolation filter coefficient is adjusted and a filter coefficient adjustment unit that generates an adjustment filter coefficient and the adjustment filter coefficient A pixel interpolation unit that derives and outputs a pixel value of each pixel of the transformed video by interpolation;
A video display device comprising: The video display device according to claim 5,
The video display device outputs video data representing video obtained by performing trapezoidal distortion correction on the original video to the video display unit.

Images