optimal_seamline.py

#!/usr/bin/python
import numpy as np
import cv2


def imgLabeling(img1, img2, img3, img4, maskSize, xoffsetL, xoffsetR,
                minloc_old=None):
    if len(img1.shape) == 3:
        errL = np.sum(np.square(img1.astype(np.float64) -
                                img2.astype(np.float64)), axis=2)
        errR = np.sum(np.square(img3.astype(np.float64) -
                                img4.astype(np.float64)), axis=2)
    else:
        errL = np.square(img1.astype(np.float64) - img2.astype(np.float64))
        errR = np.square(img3.astype(np.float64) - img4.astype(np.float64))
    EL = np.zeros(errL.shape, np.float64)
    ER = np.zeros(errR.shape, np.float64)
    EL[0] = errL[0]
    ER[0] = errR[0]
    for i in range(1, maskSize[1]):
        EL[i, 0] = errL[i, 0] + min(EL[i - 1, 0], EL[i - 1, 1])
        ER[i, 0] = errR[i, 0] + min(ER[i - 1, 0], ER[i - 1, 1])
        for j in range(1, EL.shape[1] - 1):
            EL[i, j] = errL[i, j] + \
                min(EL[i - 1, j - 1], EL[i - 1, j], EL[i - 1, j + 1])
            ER[i, j] = errR[i, j] + \
                min(ER[i - 1, j - 1], ER[i - 1, j], ER[i - 1, j + 1])
        EL[i, -1] = errL[i, -1] + min(EL[i - 1, -1], EL[i - 1, -2])
        ER[i, -1] = errR[i, -1] + min(ER[i - 1, -1], ER[i - 1, -2])
    minlocL = np.argmin(EL, axis=1) + xoffsetL
    minlocR = np.argmin(ER, axis=1) + xoffsetR
    if minloc_old is None:
        minloc_old = [minlocL, minlocR, minlocL, minlocR]
    minlocL_fin = np.int32(0.4 * minlocL + 0.3 *
                           minloc_old[0] + 0.3 * minloc_old[2])
    minlocR_fin = np.int32(0.4 * minlocR + 0.3 *
                           minloc_old[1] + 0.3 * minloc_old[3])
    mask = np.ones((maskSize[1], maskSize[0], 3), np.float64)
    for i in range(maskSize[1]):
        mask[i, minlocL_fin[i]:minlocR_fin[i]] = 0
        mask[i, minlocL_fin[i]] = 0.5
        mask[i, minlocR_fin[i]] = 0.5
    cv2.imshow('mask', mask.astype(np.float32))
    return mask, [minlocL, minlocR, minlocL_fin, minlocR_fin]