lab2 submission

'''

bicubic interpolation

11812418

My SID ends with 8, so I need to enlarge the image for 1.8x (461) and shrink the image for 0.8x (205)

'''

from scipy import interpolate

import numpy as np

from PIL import Image

def bicubic_11812418(input_file, dim):

    image = Image.open(input_file)

    imarray = np.array(image)

    # find the dimenssion of input file

    input_height = imarray.shape[0]

    input_width = imarray.shape[1]

    # set the output

    # dim(0) is the height

    # dim(1) is the width

    # init an array

    output_arr = np.zeros(dim)

    delta_height = 1

    delta_width = 1

    # base: output

    # construct the bicubic function

    def interp2d_bicubic_scipy(array_data, relative_y, relative_x):

        x = [0, 1, 2, 3]

        y = [0, 1, 2, 3]

        f = interpolate.interp2d(y, x, array_data, kind='cubic')

        interp_result = f(relative_y, relative_x)

        return interp_result

    # iterate in output array

    for i in range(dim[0]):

        for j in range(dim[1]):

            # i height (dim0)

            # j width (dim1)

            # start from 1

            # transform to input coordinate

            projected_height = (i) * (input_height - 1) / (dim[0] - 1)

            projected_width = (j) * (input_width - 1) / (dim[1] - 1)

            # print(str(projected_height) + ' ' + str(projected_width))

            # find the border

            interpolation_x_floor = int(np.floor(projected_width))

            interpolation_y_floor = int(np.floor(projected_height))

            array16 = []

            frame_position_x = interpolation_x_floor

            frame_position_y = interpolation_y_floor

            # special case

            # XY MIN

            if ((interpolation_x_floor - 1 < 0) or (interpolation_y_floor - 1 < 0) or (

                    interpolation_x_floor + 2 > input_width - 1) or (interpolation_y_floor + 2 > input_height - 1)):

                border_X_Min = max(0, interpolation_x_floor - 1)

                border_Y_Min = max(0, interpolation_y_floor - 1)

                border_X_Max = min(input_width - 1, interpolation_x_floor + 2)

                border_Y_Max = min(input_height - 1, interpolation_y_floor + 2)

                if ((interpolation_x_floor - border_X_Min) < (border_X_Max - interpolation_x_floor - 1)):

                    # left border

                    frame_position_x = border_X_Min + 1

                elif ((interpolation_x_floor - border_X_Min) > (border_X_Max - interpolation_x_floor - 1)):

                    # right border

                    frame_position_x = border_X_Max - 2

                if ((interpolation_y_floor - border_Y_Min) < (border_Y_Max - interpolation_y_floor - 1)):

                    # lower border

                    frame_position_y = border_Y_Min + 1

                elif ((interpolation_y_floor - border_Y_Min) > (border_Y_Max - interpolation_y_floor - 1)):

                    # upper border

                    frame_position_y = border_Y_Max - 2

            for k in range(4):

                array16 = np.concatenate(

                    (array16, imarray[frame_position_y - 1 + k][frame_position_x - 1:frame_position_x + 3]), axis=None)

            output_arr[i][j] = interp2d_bicubic_scipy(array16, (projected_height - frame_position_y + 1),(projected_width - frame_position_x + 1))

    return output_arr

#Testbench

#enlarge

dim1 = (461, 461)

output_img = bicubic_11812418('rice.tif', dim1)

im = Image.fromarray(output_img.astype(np.uint8))

im.save("enlarged_bicubic_11812418.tif")

#shrink

dim2 = (205, 205)

output_img = bicubic_11812418('rice.tif', dim2)

im = Image.fromarray(output_img.astype(np.uint8))

im.save("shrunk_bicubic_11812418.tif")

 No newline at end of file

'''

bilinear interpolation

11812418

My SID ends with 8, so I need to enlarge the image for 1.8x (461) and shrink the image for 0.8x (205)

'''

import numpy as np

from PIL import Image

def bilinear_11812418(input_file, dim):

    image = Image.open(input_file)

    imarray = np.array(image)

    #find the dimenssion of input file

    input_height = imarray.shape[0]

    input_width = imarray.shape[1]

    #set the output

    #dim(0) is the height

    #dim(1) is the width

    #init an array

    output_arr = np.zeros(dim)

    delta_height = 1

    delta_width = 1

    #base: output

    #iterate in output array

    for i in range(dim[0]):

        for j in range(dim[1]):

            #i height (dim0)

            #j width (dim1)

            #start from 1

            #transform to input coordinate

            projected_height = (i)*(input_height-1)/(dim[0]-1)

            projected_width = (j)*(input_width-1)/(dim[1]-1)

            #print(str(projected_height) + ' ' + str(projected_width))

            #find the border

            interpolation_h_up = int(np.ceil(projected_height))

            interpolation_h_down = int(np.floor(projected_height))

            interpolation_w_up = int(np.ceil(projected_width))

            interpolation_w_down = int(np.floor(projected_width))

            k_w_down = (abs(int(imarray[interpolation_h_up][interpolation_w_down])-int(imarray[interpolation_h_down][interpolation_w_down])))/delta_height

            w_down_val = int(imarray[interpolation_h_down][interpolation_w_down]) + k_w_down*(projected_height-interpolation_h_down)

            k_w_up = (abs(int(imarray[interpolation_h_up][interpolation_w_up])-int(imarray[interpolation_h_down][interpolation_w_up])))/delta_height

            w_up_val = int(imarray[interpolation_h_down][interpolation_w_up]) + k_w_up*(projected_height-interpolation_h_down)

            #special case

            if(projected_height % 1):

                w_down_val = imarray[int(projected_height)][int(projected_width)]

                w_up_val = imarray[int(projected_height)][int(projected_width)]

            if(projected_width % 1):

                interpolation_w_down = projected_width

            k_h = (abs(w_up_val-w_down_val)/delta_width)

            h_val = w_down_val + k_h*(projected_width-interpolation_w_down)

            output_arr[i][j] = h_val

    return output_arr

#Testbench

#enlarge

dim1 = (461, 461)

output_img = bilinear_11812418('rice.tif', dim1)

im = Image.fromarray(output_img.astype(np.uint8))

im.save("enlarged_bilinear_11812418.tif")

#shrink

dim2 = (205, 205)

output_img = bilinear_11812418('rice.tif', dim2)

im = Image.fromarray(output_img.astype(np.uint8))

im.save("shrunk_bilinear_11812418.tif")

 No newline at end of file