This can be solved algebraically as follows:
$$\frac{1}{2}-\left| y\right| =\sqrt{\left(\frac{1}{2}-\left| y\right| \right)^2+(y-\sin (x))^2} $$
Firstly note that $\frac{1}{2}-|y|\ge0$ so $-\frac{1}{2}\le y\le\frac{1}{2}$.
Next square the equation:
$$\left(\frac{1}{2}-\left| y\right|\right)^2 =\left(\frac{1}{2}-\left| y\right| \right)^2+(y-\sin (x))^2 $$
$$0=(y-\sin(x))^2$$
$$0=y-\sin(x)$$
$$y=\sin(x)$$
So the solution is $y=\sin(x)$ such that $-\frac{1}{2}\le y\le\frac{1}{2}$.
Using Mathematica's Solve or Reduce gives the same result (it just gives restrictions on $x$ rather than $y$).
Solve[{1/2 - Abs[y] == Sqrt[(1/2 - Abs[y])^2 + (y - Sin[x])^2], Element[y, Reals], Element[x, Reals]}, y]
{{y -> ConditionalExpression[Sin[x], (C[1] \[Element] Integers && 2 \[Pi] C[1] <= x <= 1/6 (\[Pi] + 12 \[Pi] C[1])) || (C[1] \[Element] Integers && \[Pi] + 2 \[Pi] C[1] < x <= 1/6 (7 \[Pi] + 12 \[Pi] C[1])) || (C[1] \[Element] Integers && 1/6 (-\[Pi] + 12 \[Pi] C[1]) <= x < 2 \[Pi] C[1]) || (C[1] \[Element] Integers && 1/6 (5 \[Pi] + 12 \[Pi] C[1]) <= x <= \[Pi] + 2 \[Pi] C[1])]}}
Reduce[{1/2 - Abs[y] == Sqrt[(1/2 - Abs[y])^2 + (y - Sin[x])^2], Element[y, Reals], Element[x, Reals]}, y]
C[1] \[Element] Integers && (((2 \[Pi] C[1] <= x <= 1/6 (\[Pi] + 12 \[Pi] C[1]) || 1/6 (5 \[Pi] + 12 \[Pi] C[1]) <= x <= \[Pi] + 2 \[Pi] C[1]) && y == Sin[ x]) || ((1/6 (-\[Pi] + 12 \[Pi] C[1]) <= x < 2 \[Pi] C[1] || \[Pi] + 2 \[Pi] C[1] < x <= 1/6 (7 \[Pi] + 12 \[Pi] C[1])) && y == Sin[x]))
Plotting the solution from the algebraic solution is then straightforward:
Plot[Sin[x] && Abs[Sin[x]] <= 1/2, {x, -Pi, Pi}]

ParametricPlot[If[1/2 - Abs[Sin[x]] > 0, {x, Sin[x]}, {}], {x, -Pi, Pi}]– Coolwater May 25 '17 at 07:57FullSimplify[Reduce[1/2 - Abs[y] == Sqrt[(1/2 - Abs[y])^2 + (y - Sin[x])^2], {x, y}, Reals], {x, y} ∈ Reals]might be helpful in making a plot. – J. M.'s missing motivation May 25 '17 at 08:29ContourPlothas a hard time plotting the contour of $f(x,y)=g(x,y)$ if the functions never cross each other, i.e. you always have $f(x,y)\le g(x,y)$ and never $f(x,y)>g(x,y)$. – May 26 '17 at 04:22