With the rapid advancement of communication technologies, the Internet of Things, and information-gathering systems, secure transmission of image data through insecure networks has become a critical necessity. Chaotic systems have attracted significant attention in image encryption due to their high sensitivity to initial conditions, pseudo-random behavior, and extremely long periodic orbits. This paper proposes a novel image encryption algorithm based on a four-dimensional chaotic system with a bounded attractor and perfect decryption capability. The proposed 4D chaotic system is designed based on the classical Chen system, and by incorporating a new differential equation, it exhibits hyperchaotic behavior with two positive Lyapunov exponents (λ₁ = 0.852 and λ₂ = 0.124). The algorithm combines chaotic sequence-based permutation operations with forward and backward diffusion processes using key-dependent coefficients. Experimental evaluations on eight standard grayscale images of size 256×256 pixels demonstrate that the proposed algorithm achieves perfect decryption with zero mean squared error (MSE = 0), an average entropy of 7.995852, key sensitivity of 99.62%, and an encryption time of 0.69 seconds. Adjacent pixel correlation analysis confirms effective elimination of spatial correlations, with absolute mean correlation coefficients below 0.03. The key space, generated using the SHA-256 hash function, is 2^256, significantly exceeding the 2^128 security threshold and ensuring resistance against brute-force attacks. The algorithm's outstanding features—perfect decryption, high speed, implementation simplicity, and computational efficiency—make it highly suitable for real-time applications in electronic warfare systems, secure military communications, and unmanned platforms.