Product image interpolating device, method and non-transitory information recording medium
A product imaging device ( 121 ) is provided which facilitates a user to capture the image sequence of entire surroundings of a product. An image sensor unit ( 201 ) senses incident light from the external world where the product is disposed and outputs an image representing a result of the sensing. An instruction receiving unit ( 202 ) receives an image-capture instruction from the user. A memory unit ( 203 ) stores the image sensed by the image sensor unit ( 201 ) upon reception of an image-capture instruction. A finder display unit ( 204 ) synthesizes the image stored in the memory unit ( 203 ) with an image presently sensed by the image sensor unit ( 201 ) and displays a synthesized image on a finder screen.
1. An image processing device comprising:
at least one non-transitory memory operable to store program code;
at least one processor operable to read said program code and operate as instructed by said program code, said program code including:
angle accepting code that causes said at least one processor to accept an angle around a rotational axis passing through a product; and
interpolating code that causes said at least one processor to interpolate an image sequence of entire surroundings of the product including a plurality of normalized images each associated with a rotational angle based on the accepted angle, and to obtain an interpolated image showing an appearance of the product as viewed at the accepted angle,
wherein the interpolating code further causes said at least one processor to:
select normalized images B[k] and B[k+1] associated with rotational angles θ[k] and θ[k+1] between the accepted angle φ among rotational angles θ[0], θ[1], . . . and θ[N−1] (where θ[0]≦θ[1]≦ . . . and ≦θ[N−1]) associated with respective normalized images B[0], B[1], . . . and B[N−1] included in the image sequence of entire surroundings,
generate a first image L(φ) L[k] −1 B[k] and a second image L(φ) L[k+1] −1 B[k+1] from affine transformation matrixes L[k] and L[k+1] for projecting a real world object in normalized images B[k] and B[k+1], and an affine transformation matrix L(φ) relative to the accepted angle φ for projecting the real world in an image,
divide the first image L(φ) L[k] −1 B[k] into image regions of (φ−θ[k]) and (θ[k+1]−φ) through a dividing line parallel to the center axis, and extract a third image at a side where the rotational angle around the center axis is smaller than the accepted angle,
divide the second image L(φ) L[k+1] −1 B[k+1] into image regions of (φ−θ[k]) and (θ[k+1]−φ) through a dividing line parallel to the center axis, and extract a fourth image at a side where the rotational angle around the center axis is larger than the accepted angle, and
dispose and arrange side by side the third image and the fourth image at a side where the rotational angle around the center axis is smaller than the accepted angle and a side where the rotational angle around the center axis is larger than the accepted angle, thereby obtaining the interpolated image.
2. An image processing method comprising:
an angle accepting step of accepting an angle around a rotational axis passing through a product; and
an interpolating step of interpolating an image sequence of entire surroundings of the product including a plurality of normalized images each associated with a rotational angle based on the accepted angle, and obtaining an interpolated image showing an appearance of the product as viewed at the accepted angle,
in the interpolating step,
selecting normalized images B[k] and B[k+1] associated with rotational angles θ[k] and θ[k+1] between the accepted angle φ among rotational angles θ[0], θ[1], . . . and θ[N−1] (where θ[0]≦θ[1]≦ . . . and ≦θ[N−1]) associated with respective normalized images B[0], B[1], . . . and B[N−1] included in the image sequence of entire surroundings,
generating a first image L(φ) L[k] −1 B[k] and a second image L(φ) L[k+1] −1 B[k+1] from affine transformation matrixes L[k] and L[k+1] for projecting a real world object in normalized images B[k] and B[k+1], and an affine transformation matrix L(φ) relative to the accepted angle φ for projecting the real world in an image,
dividing the first image L(φ) L[k] −1 B[k] into image regions of (φ−θ[k]) and (θ[k+1]−φ) through a dividing line parallel to the center axis, and extracting a third image at a side where the rotational angle around the center axis is smaller than the accepted angle,
dividing the second image L(φ) L[k+1] −1 B[k+1] into image regions of (φ−θ[k]) and (θ[k+1]−φ) through a dividing line parallel to the center axis, and extracting a fourth image at a side where the rotational angle around the center axis is larger than the accepted angle, and
disposing and arranging side by side the third image and the fourth image at a side where the rotational angle around the center axis is smaller than the accepted angle and a side where the rotational angle around the center axis is larger than the accepted angle, thereby obtaining the interpolated image.
3. A non-transitory computer-readable non-temporal information recording medium recording therein a program that allows a computer to function as:
an angle accepting unit that accepts an angle around a rotational axis passing through a product; and
an interpolating unit which interpolates an image sequence of entire surroundings of the product including a plurality of normalized images each associated with a rotational angle based on the accepted angle, and which obtains an interpolated image showing an appearance of the product as viewed at the accepted angle,
the interpolating unit,
selects normalized images B[k] and B[k+1] associated with rotational angles θ[k] and θ[k+1] between the accepted angle φ among rotational angles θ[0], θ[1], . . . and θ[N−1] (where θ[0]≦θ[1]≦ . . . and ≦θ[N−1]) associated with respective normalized images B[0], B[1], . . . and B[N−1] included in the image sequence of entire surroundings,
generates a first image L(φ) L[k] −1 B[k] and a second image L(φ) L[k+1] −1 B[k+1] from affine transformation matrixes L[k] and L[k+1] for projecting a real world object in normalized images B[k] and B[k+1], and an affine transformation matrix L(φ) relative to the accepted angle φ for projecting the real world in an image,
divides the first image L(φ) L[k] −1 B[k] into image regions of (φ−θ[k]) and (θ[k+1]−φ) through a dividing line parallel to the center axis, and extracts a third image at a side where the rotational angle around the center axis is smaller than the accepted angle,
divides the second image L(φ) L[k+1] −1 B[k+1] into image regions of (φ−θ[k]) and (θ[k+1]−φ) through a dividing line parallel to the center axis, and extracts a fourth image at a side where the rotational angle around the center axis is larger than the accepted angle, and
disposes and arranges side by side the third image and the fourth image at a side where the rotational angle around the center axis is smaller than the accepted angle and a side where the rotational angle around the center axis is larger than the accepted angle, thereby obtaining the interpolated image.