Abstract: An initial high-resolution image (D2) is generated by projection of low-resolution images (DIN) onto a high-resolution image space, an intermediate-resolution image (D31) is generated by projection of the low-resolution images (DIN) onto an intermediate-resolution image space, the intermediate-resolution image (D31) is interpolated and enlarged to generate an intermediate-resolution enlarged image (D3), and the pixel values of the undefined pixels in the initial high-resolution image (D2) are estimated using the pixel values of the corresponding pixels in the intermediate-resolution enlarged image (D3). It is possible to generate a high-resolution image of a high picture quality with higher accuracy even when less low-resolution images are used.

Abstract: An imaging signal obtained during pattern light projection and an imaging signal during pattern light non-projection are separated from an imaging signal, and the projection angle of a light spot is specified on the basis of an array of light spots in the projection image component. The projected pattern light includes a plurality of cells that accompanies the light spots and that constitutes a discrimination code, and specifies the position of the light spots accompanied by the discrimination code in the projection pattern on the basis of the discrimination code. Distance information to an object to be imaged can be acquired using a low amount of computation.

Abstract: In an image processing for correcting a distorted image obtained by photography by use of a super-wide angle optical system such as a fisheye lens or an omnidirectional mirror, to obtain an image of a perspective projection method, a composite index (Rn) combining a height on the projection sphere with a distance from the optical axis is computed (301), and a distance (Rf) from an origin in the distorted image is computed (302), using the composite index (Rn). Further, two-dimensional coordinates (p, q) in the distorted image are computed (303), using the distance (Rf) from the origin, and a pixel value in an output image is determined using a pixel at a position in the distorted image specified by the two-dimensional coordinates, or a pixel or pixels neighboring the specified position. It is possible to perform the projection from the projection sphere to the image plane, that is, the computation of the coordinates on the coordinate plane, while restricting the amount of computation.

Abstract: An image generation device includes: a projector for projecting pattern light of a near-infrared wavelength at intervals; an imager for outputting an imaging signal of an imaged image; a unit for generating a pattern light image by obtaining a difference between the imaging signal when the pattern light is projected and the imaging signal when the pattern light is not projected, and generating an ambient light image from the imaging signal when the pattern light is not projected; a determiner for determining a type of the ambient light; and a generator for estimating, from the type, a component due to light of a near-infrared wavelength in the ambient light image, and generating a visible light image by subtracting the component from the ambient light image. The determiner determines the type from a proportion of R, G, and B components in the imaged image, ambient light image, or visible light image.

Abstract: An image processing apparatus and image processing method enlarge an input image (Din) to generate a low-resolution enlarged image (D101). Depending on a result of identification of the pattern of the input image (Din), coefficient data (D108) for conversion to a high-resolution are selected, and a feature component (D102H) of a low resolution is converted to a feature component (D103H) of a high resolution. Decision as to whether or not the pattern of a local region of the input image (Din) is flat is made. If it is flat, the coefficient data are so selected that no substantial alteration is made to pixel values in a high-resolution conversion unit (103). It is possible to reduce the circuit size and the memory capacity, and to improve the noise immunity, and achieve conversion to a high resolution suitable for implementation by hardware.

Abstract: In a camera which electronically realizes panning, tilting, and zooming, on the basis of the luminance detected in each frame period (n?1), a condition of exposure for two frame periods thereafter (n+1), and digital gain for three frame periods thereafter (n+2) are set. When an instruction for changing the extracted region is received in a certain frame period (n?1), the region extracted from the image is changed three frame periods thereafter. It is possible to achieve a stable exposure control by which the luminance of the image varies little even when the region extracted from the image is switched.

Abstract: A region determination circuit (60) determines whether or not each of the pixels in an image is within a region subject to correction, in which pixels having at most a predetermined brightness level appear with a frequency equal to or less than a predetermined value. An offset level generation circuit (10) generates an offset level (Offset) on the basis of the brightness of the pixels determined to be within the region subject to correction. An offset subtraction circuit (1) subtracts the offset level (Offset) from the image signal (Yi) to generate an offset image signal (Yofst). A gain generation circuit (20) generates a gain for the offset image signal (Yofst). A gain multiplication circuit (2) multiplies the offset image signal (Yofst) by the gain to generate a corrected image signal (Ya).

Abstract: In an image processing for correcting a distorted image obtained by photography by use of a super-wide angle optical system such as a fisheye lens or an omnidirectional mirror, to obtain an image of a perspective projection method, of a composite index (Rn) combining a height on the projection sphere with a distance from the optical axis is computed (301), and a distance (Rf) from an origin in the distorted image is computed (302), using the composite index (Rn). Further, two-dimensional coordinates (p, q) in the distorted image are computed (303), using the distance (Rf) from the origin, and a pixel value in an output image is determined using a pixel at a position in the distorted image specified by the two-dimensional coordinates, or a pixel or pixels neighboring the specified position. It is possible to perform the projection from the projection sphere to the image plane, that is, the computation of the coordinates on the coordinate plane, while restricting the amount of computation.

Abstract: An imaging device has an intra-plane pixel summation unit and an inter-plane pixel summation unit. In the frame of interest, the intra-plane pixel summation unit sums the signals of the pixel of interest and selected highly correlated neighboring pixels to generate an intra-plane sensitized signal and a code indicating the pattern formed by the pixel of interest and the selected pixels. The inter-plane pixel summation unit selects pixels for summation from among the pixels positioned identically to the pixel of interest and pixels in the neighborhood of the identically positioned pixels in one or more frames neighboring the frame of interest on the basis of the agreement or disagreement of their pixel summing pattern codes and correlation of their intra-plane sensitized signals. In a Bayer array imaging device, enhanced sensitivity is obtained without causing color mixing.

Abstract: An NR coefficient (Kf) is calculated on the basis of a representative value (Pm) of a pixel of interest and a low-pass filtered value (La-Ld) nearest to the representative value (Pm) of the pixel of interest among a plurality of low-pass filtered values filtered one-dimensionally in different directions in an input image, and the pixel value of the pixel of interest and a value output by a two-dimensional low-pass filter (25) are weighted by the NR coefficient (Kf) and added (36). The representative value of the pixel of interest may be, for example, a median value obtained by filtering the current frame and a past frame in the time axis direction. The image processing device reduces noise without lowering resolution.

Abstract: For each pixel in an image (Din), a contrast correlation value (CT) is detected for peripheral areas centered around the pixel to be corrected (1), a contrast enhancement coefficient (Ken) is determined in accordance with the contrast correlation value (CT) (2), and in accordance with the enhancement coefficient (Ken), local contrast is enhanced for each pixel and an intermediate image (D3) is generated (3). When performing noise reduction (5) by smoothing the intermediate image (D3) in the time direction, the degree of noise reduction is controlled in accordance with a noise reduction coefficient (Knr) that is large where the enhancement coefficient (Ken) is large. With respect to a low-contrast image such as one captured under fog, haze or other poor weather conditions, the contrast in areas having reduced contrast is appropriately improved, and the noise that is enhanced in conjunction with contrast improvement is reduced, enabling a high quality image to be obtained.

Abstract: An image processing device (21) is provided with: a noise suppression processor (23) for creating a noise-reduced image (NRs) by executing spatial filtering on a plurality of pixels, which fall within a reference range that includes a pixel of interest, in an input image (Is); and a distortion correction processor (24) for correcting distortion by local stretching or shrinking by a magnification factor (R) corresponding to the local amount of distortion in the noise-reduced image (NRs). The reference range is dynamically changed in accordance with the magnification factor (R). Control is performed so that the reference range becomes narrower as the magnification factor (R) increases.

Abstract: An imaging signal obtained during pattern light projection and an imaging signal during pattern light non-projection are separated from an imaging signal, and the projection angle of a light spot is specified on the basis of an array of light spots in the projection image component. The projected pattern light includes a plurality of cells that accompanies the light spots and that constitutes a discrimination code, and specifies the position of the light spots accompanied by the discrimination rode in the projection pattern on the basis of the discrimination code. Distance information to an object to be imaged can be acquired using a low amount of computation.

Abstract: An image processing device 81 includes a distance information acquisition unit 35 and an image combination unit 30. Based on positions of a common subject on a plurality of images captured by a plurality of cameras 10, 11, 12, and 13 for capturing surrounding images of a vehicle, the distance information acquisition unit 35 calculates a first subject distance from a vehicle reference position O to the common subject, and calculates, based on the first subject distance, a second subject distance to a subject captured by one of the plurality of cameras 10, 11, 12, and 13. Based on the first subject distance and the second subject distance, the image combination unit 30 converts the captured images in which the plurality of cameras 10, 11, 12, and 13 are employed as viewpoints to images in which the reference position O is employed as a viewpoint, and combines the converted images on a single plane of projection.

Abstract: An NR coefficient (Kf) is calculated on the basis of a representative value (Pm) of a pixel of interest and a low-pass filtered value (La-Ld) nearest to the representative value (Pm) of the pixel of interest among a plurality of low-pass filtered values filtered one-dimensionally in different directions in an input image, and the pixel value of the pixel of interest and a value output by a two-dimensional low-pass filter (25) are weighted by the NR coefficient (Kf) and added (36). The representative value of the pixel of interest may be for example, a median value obtained by filtering the current frame and a past frame in the time axis direction. The image processing device reduces noise without lowering resolution.

Abstract: An imaging device has an intra-plane pixel summation unit and an inter-plane pixel summation unit. In the frame of interest, the intra-plane pixel summation unit sums the signals of the pixel of interest and selected highly correlated neighboring pixels to generate an intra-plane sensitized signal and a code indicating the pattern formed by the pixel of interest and the selected pixels. The inter-plane pixel summation unit selects pixels for summation from among the pixels positioned identically to the pixel of interest and pixels in the neighborhood of the identically positioned pixels in one or more frames neighboring the frame of interest on the basis of the agreement or disagreement of their pixel summing pattern codes and correlation of their intra-plane sensitized signals. In a Bayer array imaging device, enhanced sensitivity is obtained without causing color mixing.

Abstract: A region determination circuit (60) determines whether or not each of the pixels in an image is within a region subject to correction, in which pixels having at most a predetermined brightness level appear with a frequency equal to or less than a predetermined value. An offset level generation circuit (10) generates an offset level (Offset) on the basis of the brightness of the pixels determined to be within the region subject to correction. An offset subtraction circuit (1) subtracts the offset level (Offset) from the image signal (Yi) to generate an offset image signal (Yofst). A gain generation circuit (20) generates a gain for the offset image signal (Yofst). A gain multiplication circuit (2) multiplies the offset image signal (Yofst) by the gain to generate a corrected image signal (Ya).

Abstract: A pattern matching processing unit 10 binarizes input image data 1, compares a binarized pattern with a pattern provided for each of groups, and outputs either first information indicating a group to which a pattern that matches the binarized pattern belongs or second information indicating no match. A 0-degree direction dedicated filter 21, a 45-degree direction dedicated filter 22, a 90-degree direction dedicated filter 23, and a 135-degree direction dedicated filter 24 which are disposed correspondingly to the groups carry out smoothing processes according to the direction of an edge of the image, respectively. When the pattern matching processing unit outputs the first information, a selector 30 selects the output of either one of the filters corresponding to the group as output image data 2, whereas when the pattern matching processing unit outputs the second information, the selector outputs the input image data 1 as the output image data 2.

Abstract: An image processing device (21) is provided with: a noise suppression processor (23) for creating a noise-reduced image (NRs) by executing spatial filtering on a plurality of pixels, which fall within a reference range that includes a pixel of interest, in an input image (Is); and a distortion correction processor (24) for correcting distortion by local stretching or shrinking by a magnification factor (R) corresponding to the local amount of distortion in the noise-reduced image (NRs). The reference range is dynamically changed in accordance with the magnification factor (R). Control is performed so that the reference range becomes narrower as the magnification factor (R) increases.

Abstract: For each pixel in an image (Din), a contrast correlation value (CT) is detected for peripheral areas centered around the pixel to be corrected (1), a contrast enhancement coefficient (Ken) is determined in accordance with the contrast correlation value (CT) (2), and in accordance with the enhancement coefficient (Ken), local contrast is enhanced for each pixel and an intermediate image (D3) is generated (3). When performing noise reduction (5) by smoothing the intermediate image (D3) in the time direction, the degree of noise reduction is controlled in accordance with a noise reduction coefficient (Knr) that is large where the enhancement coefficient (Ken) is large. With respect to a low-contrast image such as one captured under fog, haze or other poor weather conditions, the contrast in areas having reduced contrast is appropriately improved, and the noise that is enhanced in conjunction with contrast improvement is reduced, enabling a high quality image to be obtained.